A Prospective Comparison of Digital Image Analysis and Routine Cytology for the Identification of Malignancy in Biliary Tract Strictures

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2004;2: A Prospective Comparison of Digital Image Analysis and Routine Cytology for the Identification of Malignancy in Biliary Tract Strictures TODD H. BARON,* GAVIN C. HAREWOOD,* ASHWIN RUMALLA,* NICOLE L. POCHRON,* LINDA M. STADHEIM,* GREGORY J. GORES,* TERRY M. THERNEAU, PIET C. DE GROEN,* THOMAS J. SEBO, DIVA R. SALOMAO, and BENJAMIN R. KIPP Divisions of *Gastroenterology and Hepatology; Pathology; and Biostatistics, Mayo Clinic, Rochester, Minnesota Background & Aims: Digital image analysis (DIA)allows quantification of nuclear DNA content and may help distinguish benign and malignant strictures of the biliary tract. Methods: One hundred ten consecutive patients undergoing endoscopic retrograde cholangiography for suspicious biliary tract strictures were enrolled in a prospective study comparing the accuracy of DIA and routine cytology (RC). Standard brush cytology sampling was performed twice by using 2 cytology brushes per patient. Both brushes were fixed in a single-specimen vial. Each specimen was formed into 1 pellet, and the sample was equally divided for evaluation by DIA and RC. DNA histograms were generated for ploidy analysis. The DIA criterion for malignancy was demonstration of aneuploidy. Results: Two patients had inadequate samples obtained for DIA analysis, 7 benign patients were excluded because of inadequate follow-up of less than 75 days, and 1 patient was lost to follow-up to clarify malignant versus benign disease. Of the remaining 100 patients, 56 strictures were malignant and 44 were benign. The sensitivities of DIA and RC were 39.3% and 17.9%, respectively (P 0.014). The specificities of DIA and RC were 77.3% and 97.7%, respectively (P 0.003). The accuracy of DIA (56.0%) was equivalent to RC (53.0%). Conclusions: DIA is a valuable adjunct to RC for detecting malignant strictures of the biliary tract. In patients with biliary tract strictures, the presence or absence of malignancy is difficult to confirm. Tumors that affect the biliary tract are often fibrotic and may not invade the ductal epithelium, especially when arising from surrounding structures or metastasizing from distant sites. As a result, the intraluminal tissue-sampling techniques used during endoscopic retrograde cholangiopancreatography (ERCP) commonly produce a poor cellular specimen that is nondiagnostic or falsely negative. Although routine brush cytology (RC) during ERCP has a high specificity, the sensitivity rates are highly variable In a recent review of the RC literature, 13 the sensitivity of RC for detection of malignancy in patients with cholangiocarcinoma ranges from 44% to 80% (mean 62%) and with known pancreatic cancer ranges from 15% to 65% (mean 37%). The most recent results from a prospective study found an overall sensitivity of only 27.7% for a single brush cytology sample obtained after balloon dilation. 14 Intraluminal forceps biopsy may increase the sensitivity by an additional 20%, 1 but the technique often requires performing a biliary sphincterotomy. Because biopsy forceps are not wire guided, they only allow biopsies to be taken from the inferior margin of a stricture, which may predispose to sampling error. Moreover, the interpretation of standard tissue-sampling methods is subjective and highly dependent on the skill of the cytopathologist. Quantification of cellular DNA content, a more objective measurement, may help identify malignant lesions of the biliary tract. Flow cytometry can be used to identify aneuploidy in brush cytology samples, but the technique requires a highly cellular sample with a large proportion of malignant cells. 6,15 A relatively newer technique, digital image analysis (DIA), can be used to determine the DNA content of individual cells and is more applicable for specimens with limited cellularity. 16 In DIA, a video camera captures the light transmitted through a glass slide specimen and converts the absorption values into pixels of variable color (white, gray, and black). Computer analysis of the pixels produces a digital image of the nucleus and other cellular constituents. Quantification of DNA content, chromatin distribution, and nuclear morphology can be determined and may suggest features of malignancy. We have previously pub- Abbreviations used in this paper: DIA, digital image analysis; ERCP, endoscopic retrograde cholangiopancreatography; PDI, peak DNA indice; RC, routine cytology by the American Gastroenterological Association /04/$30.00 PII: /S (04)

2 March 2004 DIGITAL IMAGE ANALYSIS AND ROUTINE CYTOLOGY 215 lished retrospective results using DIA for the diagnosis of biliary tract strictures. 17 This prompted us to conduct a prospective study comparing the accuracy of DIA and RC for the evaluation of biliary tract strictures of diverse etiologies. Methods After approval from our institutional review board, 110 consecutive patients undergoing ERCP for suspicious biliary tract strictures were enrolled between June 2000 and December Patients were enrolled if there was clinical suspicion of a biliary tract stricture based on symptoms, cholestatic laboratory values, or radiographic studies. Patients with primary sclerosing cholangitis and clinical deterioration from biliary obstruction were included. All patients were required to have cholangiographic documentation of a biliary stricture defined as a narrowed segment of the bile duct with a luminal diameter of 75% or less than the normal bile duct diameter. Dilation of the upstream biliary tree was also required for participation in the study. Patients with strictures that were inaccessible to brush cytology sampling were excluded. Biliary Tract Samples To produce an equivalent specimen for RC and DIA, we used a pooled sampling technique. Two separate samples were taken from the strictures using 2 separate standard cytology brushes (DLB or DLB ; Wilson-Cook, Winston-Salem, NC). The brush was passed across the lesion using 5 8 to-and-fro movements. The samples were combined and fixed in a single vial containing 20 ml of PreserveCyt solution. Two aliquots were made from the pooled samples, and each used to make a Thin Prep specimen (Thin Prep 2000; Cytyc Corp., Boxborough, MA). One Thin Prep specimen was submitted for Papanicolaou staining and routine cytological examination. The second Thin Prep specimen was used for DIA. Direct brushing and immediate Pap smear staining for RC was not done because this method entails air drying the specimen. This may introduce artifact that has been reported to produce false-negative interpretation. 18 DIA Specimens designated for DIA were stained with the Feulgen dye. After hydrolysis of nuclear DNA with hydrochloric acid, the Feulgen dye stoichiometrically binds to nucleic acids. The staining density as detected by DIA is therefore proportional to the DNA content of a specific cell. 19 Measurement of DNA content was standardized by using rat hepatocytes (Bacus Laboratories, Inc, Lombard, IL) as external controls. Up to 50 nuclei from each biliary tract brushing specimen were examined under a dark green optical filter (540 nm) and an X40 objective using an image analyzer (CAS 200, Bacus Laboratories, Inc, Lombard, IL). DNA histograms were generated for ploidy analysis automatically by plotting individual nuclear DNA content (x-axis) Figure 1. (A) Representative DIA histogram of a normal patient. DNA histograms showing cell distributions based on nuclear DNA content. 2C represents cells in the diploid range while 4C indicates tetraploid cells. Cells falling between 2C and 4C are considered to be aneuploid. (B) The brush cytology specimen demonstrates normal biliary epithelium sampled from the same patient as Figure 1A. against number of nuclei (y-axis) (Figures 1A, 2A, and 3A). DNA histograms were classified as either diploid, tetraploid, or aneuploid. DNA ploidy status was assigned to the cells based on evaluation of the DNA histogram generated by the Quantitative DNA Analysis program (Bacus Laboratories, Lombard, IL). The captured nuclei were automatically placed into 1 of 4 categories, which identified nuclei with DNA indices between 0.95 and 1.05 (diploid), 1.06 and 1.89 (Sphase or aneuploid), 1.90 and 2.10 (tetraploid), or greater than 2.10 (hypertetraploid). Diploid specimens had peak DNA indices (PDIs) between 0.95 and 1.05 without a clonal population of cells beyond This corresponded to a relatively high percentage of nuclei in the diploid category and a relatively low percentage of nuclei in the S-phase and tetraploid categories. Aneuploid specimens had PDI between 1.06 and 1.89 or greater than 2.10, and tetraploid tumors had PDI between 1.90 and Figure 2 shows representative malignant (aneuploid) and benign (diploid) histograms. The cytotechnologists performing DIA and pathologists interpreting the DNA histograms were blinded to the patients clinical

3 216 BARON ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No. 3 Classification of Strictures Strictures were classified as benign on the basis of surgical confirmation, a cancer-free clinical course for at least 6 months. Strictures were considered malignant if confirmed by surgical histology or preoperative biopsy. Patients without histological confirmation were also considered to have malignant lesions if they had progressive malignant appearing disease on subsequent imaging studies. Standardized telephone interviews were conducted monthly for 3 months and every 3 months thereafter of patients without histological confirmation to determine the final designation of their strictures (benign or malignant). Statistical Analysis The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of DIA and RC were calculated along with 95% confidence intervals. McNemar s test was used to compare the 2 methods. Based on our pilot study that showed a DIA sensitivity of 85% 17 and the near- Figure 2. (A) DNA histogram displaying cells obtained from a patient with a malignant biliary tract stricture caused by primary cholangiocarcinoma. (B) Representative brush cytology specimen from same patient as in 2A showing malignant cells. history and results of RC. DIA specimens were processed and analyzed in a routine manner with all other clinical cases performed on a daily basis in the Mayo Clinic s Image Morphometry Laboratory. For the purposes of this study, nondiploid (tetraploid and aneuploid) specimens were classified as malignant. Routine Cytology Routine cytology was interpreted without knowledge of DIA results. Interpretation for initial clinical care was rendered by the cytopathologist assigned to cover routine clinical cytology specimens for a given day who had the option to access clinical data. Routine cytology specimens were reported as equivocal, no abnormal cells (negative), and malignant cells (positive) (Figures 1B, 2B, and 3B). For the final results of the study, the specimens were blindly re-reviewed by 2 experienced cytopathologists. When the results between the 2 cytopathologists differed or were equivocal, a consensus was reached by including the initial unblinded pathology interpretation. Thus, the final results of cytology for the purposes of our analysis were either positive or negative. Sensitivities were calculated for each classification scheme. Figure 3. (A) DNA histogram displaying cells obtained from a patient with a malignant biliary tract stricture caused by pancreatic adenocarcinoma. (B) Representative brush cytology specimen from same patient as in 3A showing malignant cells.

4 March 2004 DIGITAL IMAGE ANALYSIS AND ROUTINE CYTOLOGY 217 Table 1. Performance Characteristics of Digital Image Analysis and Routine Cytology Test Sensitivity Specificity PPV NPV Accuracy DIA 39.3% a ( %) 77.3% b ( %) 68.8% ( %) 50.0% ( %) 56.0% ( %) RC 17.9% a ( %) 97.7% ( %) 90.9% ( %) 48.3% ( %) 53.0% ( %) DIA RC 42.9% ( %) 77.3% ( %) 70.6% ( %) 51.5% ( %) 58.0% ( %) NOTE. Data in parentheses are 95% confidence intervals. DIA, digital image analysis; RC, routine cytology; PPV, positive predictive value; NPV, negative predictive value. a P for sensitivity DIA vs. RC. b P for specificity RC vs. DIA. maximum literature rate of 65% for RC, a sample size of 100 patients (40 malignant, 60 benign) provided greater than 90% power (at alpha 0.05) for detecting a statistically significant difference in the sensitivity between DIA and RC. Results A total of 100 patients with bile duct strictures were evaluated by DIA and RC. Fifty-six strictures were malignant. Histological confirmation was obtained in 69.6% (39/56) of patients with malignant strictures. Of the 17 patients who did not have a tissue diagnosis, 16 died of disease progression, including 1 patient with cholangiocarcinoma who died while awaiting liver transplantation and 1 who had tissue confirmation of malignancy by tissue obtained at a subsequent ERCP. The mean follow-up of these 17 patients was 7.7 months. Thirty-three patients had cholangiocarcinoma, 14 had pancreatic adenocarcinoma, and 9 had other malignancies. Benign strictures were caused by primary sclerosing cholangitis in 40 patients. Of the remaining 4 cases, 3 were idiopathic and 1 was because of chronic pancreatitis. The mean duration of follow-up for patients with benign strictures was 16.0 months. The sensitivities of DIA and RC were 39.3% and 17.9%, respectively (P 0.014, Table 1). The specificities of DIA and RC were 77.3% and 97.7%, respectively (P 0.003). The accuracy of DIA (56.0%) and RC (53.0%) were equivalent. Routine cytology identified only 10 of 56 malignant strictures; 8 of which were also diagnosed by DIA. In 46 patients with malignant strictures and negative RC, DIA identified 14 malignancies. Overall, DIA and RC had a combined sensitivity of 42.9%. There was no appreciable difference in detection rates by DIA for the various causes of malignant strictures (Table 2). In contrast, the 10 positive diagnoses made by RC were in 7 patients who had cholangiocarcinoma and in 3 patients who had pancreatic carcinoma. RC did not identify any of the tumors that arose from structures surrounding the bile duct or metastasizing from other sites. All 10 patients with benign strictures who were falsely diagnosed as having malignancy by DIA had primary sclerosing cholangitis. The false-positive rate for DIA was 10 of 44 (22.7%). One false-positive RC interpretation occurred in a patient with PSC with benign disease (with 7.1 months of follow-up) in whom both blinded pathologists re-reviewed the specimen as being malignant. Initial pathology interpretation had been benign. Therefore, the false-positive rate of routine cytology (RC) was 1 of 44 (2.3%). Discussion This study is the largest to date examining the role of DIA in the evaluation of biliary tract strictures. In this prospective comparison, the sensitivity of DIA was significantly better than RC. DIA identified malignant biliary strictures from a variety of causes. Despite the high proportion of benign inflammatory lesions included in this study, DIA maintained a reasonably good degree of specificity. In a retrospective study conducted at the M.D. Anderson Cancer Center, Krishnamurthy et al. 20 evaluated 49 bile duct brushings with DIA. Papanicolaou-stained slides were destained and subjected to the Feulgen reaction and ploidy analysis by DIA. The finding of hyperploidy had a sensitivity of 62% and a specificity of 91%. The sensitivity of RC was 18% when atypical cells were not considered and 44% when atypical cells were included. These results are similar to what was found in our study. Yeaton et al. 21 devised a scoring system using DIA on 48 brush cytology specimens that distinguished benign biliary epithelium from malignant epithelium with 95% Table 2. Positive Sampling Frequency of Digital Image Analysis and Routine Cytology for the Causes of Malignant Biliary Obstruction Test Biliary (N 33) Pancreatic (N 14) Other (N 9) DIA 42.4% 35.7% 33.3% RC 21.2% 21.4% 0

5 218 BARON ET AL. CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 2, No. 3 sensitivity and 100% specificity. The scoring system included assessment of DNA histograms, overall DNA content, and a cell proliferation index. Their results were not compared with findings on RC. In our study, we used DIA for identifying DNA content but not cell proliferation indices or other features such as nuclear size, nuclear shape, nucleocytoplasmic ratio, or chromatin distribution. These features may improve the results of DIA and explain the different results obtained in our study. The specificity of DIA in our study was 77.3%. In a similarly designed study that evaluated 51 pancreaticobiliary strictures, flow cytometry had a sensitivity of 42% and a specificity of 77%. 6 The occurrence of false positives with flow cytometry was attributed to cellular debris, which may produce false peaks in the DNA histogram. Our results suggest that the overall specificity of DIA is similar to that of flow cytometry, although our study included patients with primary sclerosing cholangitis. The 10 false positives with DIA all occurred in patients with primary sclerosing cholangitis. Although inflammatory lesions of the biliary tract would not be expected to produce aneuploid cell populations, other parameters assessed by DIA may be required to better distinguish primary sclerosing cholangitis and carcinoma. Evaluation of nuclear area and the topographic distribution of nuclear chromatin were previously shown in a multivariate analysis to be important factors in making this distinction. 22 Because abnormal and normal values for nuclear morphologic features are not well established, we elected not to evaluate these parameters in this prospective study. However, we are evaluating the stored quantifiable nuclear morphometric features of all cell nuclei analyzed in this study to determine what, if any, role nuclear morphometry plays in the identification of malignant strictures. The sensitivity of RC in our study for detecting malignant strictures was 17.9%. The low sensitivity of RC rate in this study as compared with what has been reported in the cumulative literature 13 may be caused by several factors. Firstly, the patients referred to our tertiary care institution likely reflect a population of patients that are inherently more difficult to classify as malignant (selection bias). Secondly, we evaluated many patients with primary sclerosing cholangitis. The biliary brush cytological findings in this population of patients are difficult to interpret for malignancy. 23 We believe the most important factor, however, is the high number of PSC patients. Although DIA improved the sensitivity rate, the specificity rate in the PSC group decreased substantially. In this population, therefore, other clinical and laboratory parameters (such as serum CA 19-9) may need to be used to help better define the nature of the stricture in these patients. We believe that whatever factors may have accounted for the low rate of positivity of brush cytology may have also decreased the sensitivity of DIA, proportionally decreasing the value of both. Finally, the sensitivity rate for routine cytology in our study is not substantially different from that found in a recent prospective study in which routine cytology had a sensitivity of 27.7% after balloon dilation of the stricture. 14 In that study, both clearly positive cytologic findings as well as suspicious cytologic findings were considered positive, thus increasing the sensitivity, as compared with our study where suspicious cytology was not assumed to be malignant. We applied DIA in the clinical setting of bile duct strictures on the premise that aneuploidy and tetraploidy were equivalent to malignancy and that a diploid result meant a benign stricture. However, previously published studies have shown that ploidy status is not always diagnostic. Although abnormal DNA content is usually indicative of malignancy, premalignant lesions elsewhere in the gastrointestinal tract, such as colonic adenomas, have also been shown to show aneuploidy. 24,25 Furthermore, Brunt and Kraemer 26 identified aneuploidy by DIA in 19 of 22 (86%) paraffin-embedded malignant bile duct specimens. Consequently, assessment of DNA content alone would not identify malignant lesions that are diploid. Although we performed DIA on fresh brushing samples, previous studies have shown that DIA can be performed on stored Papanicolaou-stained specimens without compromising the accuracy of the test. 17,27 Because the cost of DIA is approximately 100 U.S. dollars per sample and the analysis takes 1 day to complete, DIA is applicable to the clinical setting. In summary, we believe DIA is a valuable adjunct and complement to routine biliary brush cytology because of the incremental sensitivity in detecting malignant biliary strictures. Further studies using DIA features other than DNA content may further improve its sensitivity, specificity, and accuracy. References 1. Jailwala J, Fogel EL, Sherman S, Gottlieb K, Flueckiger J, Bucksot LG, Lehman GA. Triple-tissue sampling at ERCP in malignant biliary obstruction. Gastrointest Endosc 2000;51: Sugiyama M, Atomi Y, Wada N, Kuroda A, Muto T. Endoscopic transpapillary bile duct biopsy without sphincterotomy for diagnosing biliary strictures: prospective comparative study with bile and brush cytology. Am J Gastroenterol 1996;91: Pugliese V, Conio M, Nicolo G, Saccomanno S, Gatteschi B. Endoscopic retrograde forceps biopsy and brush cytology of biliary strictures: a prospective study. Gastrointest Endosc 1995; 42:

6 March 2004 DIGITAL IMAGE ANALYSIS AND ROUTINE CYTOLOGY Ponchon T, Gagnon P, Berger F, Labadi M, Liaras A, Chavaillon A, Bory R. Value of endobiliary brush cytology and biopsies for the diagnosis of malignant bile duct stenosis: results of a prospective study. Gastrointest Endosc 1995;42: Lee JG, Leung JW, Baillie J, Layfield LJ, Cotton PB. Benign, dysplastic, or malignant-making sense of endoscopic bile duct brush cytology: results in 149 consecutive patients. Am J Gastroenterol 1995;90: Ryan ME, Baldauf MC. Comparison of flow cytometry for DNA content and brush cytology for detection of malignancy in pancreaticobiliary strictures. Gastrointest Endosc 1994;40: Ferrari AP, Lichtenstein DR, Slivka A, Chang C, Carr-Locke DL. Brush cytology during ERCP for the diagnosis of biliary and pancreatic malignancies. Gastrointest Endosc 1994;40: Kurzawinski TR, Deery A, Dooley JS, Dick R, Hobbs KEF, Davidson BR. A prospective study of biliary cytology in 100 patients with bile duct strictures. Hepatology 1993;18: Ryan M. Cytologic brushings of ductal lesions during ERCP. Gastrointest Endosc 1991;37: Foutch PG, Kerr DM, Harlan JR, Kummet TD. A prospective, controlled analysis of endoscopic cytotechniques for diagnosis of malignant biliary strictures. Am J Gastroenterol 1991;86: Venu RP, Geenen JE, Kini M, Hogan WJ, Payne M, Johnson GK, Schmalz MJ. Endoscopic retrograde brush cytology. A new technique. Gastroenterology 1990;99: Scudera PL, Koizumi J, Jacobson IM. Brush cytology evaluation of lesions encountered during ERCP. Gastrointest Endosc 1990;36: De Bellis M, Sherman S, Fogel EL, Cramer H, Chappo J, McHenry L Jr, Watkins JL, Lehman GA. Tissue sampling at ERCP in suspected malignant biliary strictures (part 1). Gastrointest Endosc 2002;56: De Bellis M, Fogel EL, Sherman S, Watkins JL, Chappo J, Younger C, Cramer H, Lehman GA. Influence of stricture dilation and repeat brushing on the cancer detection rate of brush cytology in the evaluation of malignant biliary obstruction. Gastrointest Endosc 2003;58: So MJ, Cheville JC, Katzmann JA, Riehle DL, Lohse CM, Pankratz VS, Sebo TJ. Factors that influence the measurement of prostate cancer DNA ploidy and proliferation in paraffin embedded tissue evaluated by flow cytometry. Mod Pathol 2001;14: Sebo TJ. Digital image analysis. Mayo Clin Proc 1995;70: Rumalla A, Baron TH, Leontovich O, Burgart LJ, Yacavone RF, Therneau TM, de Groen PC, Sebo TJ. Improved diagnostic yield of endoscopic biliary brush cytology by digital image analysis. Mayo Clin Proc 2001;76: Logrono R, Kurtycz DF, Molina CP, Trivedi VA, Wong JY, Block KP. Analysis of false-negative diagnoses on endoscopic brush cytology of biliary and pancreatic duct strictures: the experience at 2 university hospitals. Arch Pathol Lab Med 2000;124: Kiss R, Salmon I, Camby I, Gras S, Pasteels JL. Characterization of factors in routine laboratory protocols that significantly influence the Feulgen reaction. J Histochem Cytochem 1993;41: Krishnamurthy S, Katz RL, Shumate A, Strohlein K, Khanna A, Tucker SL, Raijman I, Lahoti S. DNA image analysis combined with routine cytology improves diagnostic sensitivity of common bile duct brushing. Cancer 2001;93: Yeaton P, Kiss R, Deviere J, Salmon I, Bourgeois N, Pasteels JL, Cremer M. Use of cell image analysis in the detection of cancer from specimens obtained during endoscopic retrograde cholangiopancreatography. Am J Clin Pathol 1993;100: Sears RJ, Duckworth CW, Decaestecker C, Bourgeois N, Ledent T, Deviere J, Salmon I, Kiss R, Yeaton P. Image cytometry as a discriminatory tool for cytologic specimens obtained by endoscopic retrograde cholangiopancreatography. Cancer 1998;84: Lindberg B, Arnelo U, Bergquist A, Thorne A, Hjerpe A, Granqvist S, Hansson LO, Tribukait B, Persson B, Broome U. Diagnosis of biliary strictures in conjunction with endoscopic retrograde cholangiopancreaticography, with special reference to patients with primary sclerosing cholangitis. Endoscopy 2002;34: van den Ingh HF, Griffioen G, Cornelisse CJ. DNA aneuploidy in colorectal adenomas. Br J Cancer 1987;55: Banner BF, Chacho MS, Roseman DL, Coon JS. Multiparameter flow cytometric analysis of colon polyps. Am J Clin Pathol 1987; 87: Brunt EM, Kraemer BB. DNA image analysis study of lesions of the gallbladder and biliary system. Liver Transplant Surg 1996; 2: Sidoni A, Cavaliere A, Alunno P, Bucciarelli E. DNA-ploidy studies on cytological preparations from breast cancers by image analysis: comparison with Feulgen staining performed in destained Papanicolaou slides. Cytometry 1996;26: Address requests for reprints to: Todd H. Baron, M.D., F.A.C.P., Mayo Clinic Rochester, 200 First Street SW, Rochester, Minnesota baron.todd@mayo.edu; fax: (507) The authors thank the DIA technologists of Mayo Clinic s Image Morphometry Laboratory for their work in completing this study.