Printer Friendly

Preservation of urine for flow cytometric and visual microscopic testing.

Preservation of formed elements in urine has gained renewed interest since the development of analytical systems that allow automated analysis of erythrocytes, leukocytes, and bacteria in urine (1,2). If not preserved, urine particles should be examined within 1 h after voiding at ambient temperature or within 4 h if refrigerated to avoid lysis of material (3,4). Urine leukocyte counts in specimens from children may be questionable after 2-4 h even with refrigeration (5). Traditionally, ethanol (500 mL/L) is used to preserve uroepithelial cells for cytology. Polyethylene glycol (20 g/L) has been added to the ethanol fixative (Saccomanno's fixative) to improve preservation results (6). Alternative fixatives have also been reported (7). Preservatives have traditionally been used in many countries for specimens requiring microbiologic investigation, i.e., urine bacterial culture (8). Containers prefilled with boric acid alone (9) or in combination with formic acid or other stabilizing media have been used (10, 11). We investigated ways to preserve urine particles for automated analysis with the UF-100TM and for visual microscopy, using both generic and commercial approaches.

Materials and Methods

PATIENT URINES AND PRESERVATION PROCEDURES

Consecutive morning urine specimens were collected from adult inpatients as a part of their clinical examination at Tampere University Hospital. Test strip measurements (Combur-10 Test M and Miditron M; Roche Diagnostics) were used to select 106 specimens positive for leukocyte esterase, hemoglobin, albumin, or nitrite to enrich the fraction of pathologic findings. Aliquots of 10 mL were taken for five different storage procedures: refrigeration (4 [degrees]C) without preservatives (procedure 1); preservation of 10-mL specimens in Urine C&S tubes with lyophilized borate-formate-sorbitol powder (BD Preanalytical Solutions; currently claimed for storage of specimens for urine bacterial culture) at 20 [degrees]C (procedure 2); fixation of 9 mL of urine with 1 mL of 100 mL/L formaldehyde supplemented with 1.5 mol/L NaCl at 20 [degrees]C (procedure 3); fixation of 9 mL of urine with 1 mL of 800 mL/L ethanol supplemented with 200 g/L polyethylene glycol at 20 [degrees]C (procedure 4); and storing the urine specimen without preservatives at 20 [degrees]C (background control; procedure 5).

PARTICLE ANALYSIS

Automated urinalysis was performed by urine flow cytometry (Sysmex UF-100; SYSMEX-EUROPE GmbH) in the manual mode for sample introduction (to avoid lysis of cells caused by the mixing device) (1,2,12). In bacterial counting, the earlier software with two separate channels was used (BACT channel, counting particles with a diameter [greater than or equal to]2 [micro]m; and HIGH BACT channel, counting particles [greater than or equal to]0.6 [micro]m). In the results, only counts from the BACT channel are reported. The following time points were chosen to simulate daily workflow: (a) the morning of arrival in the laboratory (day 0, am); (b) the same day in the afternoon (day 0, pm); (c) the following day (day 1); and (d) the 3rd day after the day of specimen arrival to simulate a weekend delay (day 3). Data were collected online and processed with Microsoft Excel spreadsheets (Microsoft Corporation).

In addition, visual microscopic analysis was performed on 200-[micro]L aliquots of noncentrifuged specimens. After 20 [micro]L of Alcian blue-pyronin B supravital stain was added to the urine (13), 1-[micro]L volumes were counted in disposable chambers (Fast-Read 10; Bio-Sigma) by both brightfield and phase-contrast microscopy (Nikon Eclipse E400; Nikon Europe B.V.). Specimens were counted first on arrival, and final preservation was evaluated after storage for 3 days, using the following ordinal scale: A = no destruction of particles; B = destruction detected but classification possible; C = particles detected but difficult to classify; and D = no particles detected. Successful preservation was defined to include categories A and B at day 3.

STATISTICAL ANALYSIS

Statistics focused on comparing the stability of particle counts in repeated measurements of each tube within each preservation procedure. The Box-Cox transformation was applied to nonnormally distributed results. The following statistical model was used to analyze the variance of serial measurements: y ~ Specimen + procedure + time point + (procedure x time point) + error. The main item of interest was the interaction term (procedure x time point), which, if nonsignificant, indicated that the specimens were preserved within a given procedure over different time points. If the term was significant, a pair-wise comparison with the original counts of that procedure was performed using Tukey simultaneous tests.

To determine the degree of clinically significant bidirectional changes (from positive to negative, and simultaneously from negative to positive) during the preservation time, results were also analyzed by grouping them into three categories at each clinical cutoff limit. A gray zone of [+ or -]20% at the cutoff count was chosen to visualize counting imprecision, which is high because of generally low particle concentrations in urine. Using Microsoft Excel spreadsheets, we calculated the Cohen [kappa] coefficient to subtract random agreement from paired observations (14). The [kappa] coefficient varies between -1 (total disagreement) and + 1 (complete agreement, complete stability); a value of 0 indicates no agreement (no stability). A coefficient of 0.8 (80% agreement from all obtainable nonrandom disagreement) may be considered as a good, and 0.6 as minimum for a clinical laboratory test (15). An exact test was used to calculate the statistical significance of each obtained [kappa] coefficient (null hypothesis is random distribution; [kappa] = 0).

Results

Fourteen of 106 specimens (13%) were rejected in flow cytometric analysis because of a high total count, producing an instrument-error message. In analysis of red blood cells (RBCs), [4] an additional three to seven specimens were rejected with different preservation procedures because of difficulties in instrumental classification (Discrimination error RBC). Quantitative counts were obtained for bacteria (BACT channel, counting particles with diameter [greater than or equal to] 2 [micro]m), RBCs, white blood cells (WBCs), casts (CAST channel; gated to detect mainly inclusion-bearing granular or cellular casts because hyaline casts are difficult to discern from slime threads with such an analyzer), large epithelial cells (ECs), and small round cells (SRCs; including both epithelial and inflammatory cells).

PRESERVATION OF INITIAL PARTICLE CONCENTRATIONS

After the initial counting (day 0, am), follow-up counts were measured on the same day in the afternoon (day 0, pm), on the following day (day 1) and 3 days later (day 3). The original median and mean values and the ranges for all particle counts are shown in Table 1. An increment of BACT counts was seen in tubes stored without preservatives at 20 [degrees]C (procedure 5) for 1 and 3 days. Bacterial counts were also increased in procedure 4 (ethanolpolyethylene glycol) at 3 days (Table 1). RBC counts tended to decrease during storage; a statistically significant reduction in counts was seen in RBCs fixed with formaldehyde solution (procedure 3) at 1 and 3 days (Table 1). Somewhat surprisingly, the mean RBC counts did not decrease at room temperature (procedure 5). Also unexpectedly, WBCs did not show any differences during the follow-up period. Casts and large ECs values were artifactually higher with procedure 3 after 1 or 3 days (Table 1). Cast counts were higher in procedure 4 than in the other procedures at all time points, but they remained stable during the observation period. Finally, SRC counts showed no significant changes in the mean particle concentrations during follow-up.

BIDIRECTIONAL CHANGES IN PARTICLE CONCENTRATIONS

During a preservation period, increases and decreases in particle counts may occur, both of which may be clinically significant. These changes may not alter the mean counts for particles at given time points because alterations in the opposite direction may compensate each other. Lysis of RBCs or WBCs creates false-negative (FN) results. Increases in counts may result from bacterial growth, precipitation of new particulate material, or misclassification caused by in vitro changes in particle size or granularity, leading to false-positive (FP) results. Therefore, cross-tables were created to demonstrate changes in counts for individual specimens. Results were divided into positive and negative subgroups for each particle at chosen clinical cutoff concentrations. A gray zone of [+ or -] 20% at each cutoff count was applied to allow for the analytical imprecision of counting.

Shown in Fig. 1 is an example of RBC counts obtained using preservation procedure 2 (BD tube). The original counts (Day 0, am) form the horizontal grouping shown in columns, whereas each of the follow-up counts (Day 0, pm; Day 1; and Day 3) form the rows of cross-tables. For RBCs, a cutoff gray zone of 8-12 x [10.sup.6] RBCs/L was used. The stability of particle counts is shown diagonally, with positive, borderline, and negative specimens (shown in bold font) remaining in their original categories in the follow-up. Stability remained uncertain in gray zone counts (6 + 3 + 3 + 3 = 15 specimens) because of statistical imprecision: the UF-100 flow cytometer counts ~8 [micro]L of urine, the total count being distributed according to Poisson distribution. FPs consisted of specimens that in the follow-up showed artifactually increased counts; the left lower corner was included in the FPs only. FNs in the follow-up comprised specimens that lost their counts (the upper right corner included only). On day 3, an increase in FP specimens to 4.8% was seen, because of difficulty in discerning RBCs from other small particles (bacteria, noncellular debris, and precipitate). The Cohen [kappa] coefficient expresses the fraction of nonrandom stability (agreement) for the diagonal (Fig. 1).

All [kappa] coefficients observed for all classified elements with each preservation procedure were graded as summarized in Table 2. The original detailed data are available in an online data supplement available at the Clinical Chemistry Online website (http://www.clinchem.org/content/vol48/issue6). In bacterial counts (BACT gray zone, 160-240 x [10.sup.6]/L), a deterioration in agreement was seen, as expected, for procedure 5 (no preservatives at 20 [degrees]C) after 3 days: agreement was only 56%; the FP rate increased to 28.1%; and the [kappa] coefficient was reduced to 0.31 (95% confidence interval, 0.15-0.47). This preservation was rated as "Poor". Procedure 1 maintained 91% agreement with the original BACT counts for 3 days with [kappa] = 0.86 and no FP or FN results. Good agreement in BACT counts was also seen with procedures 2 ([kappa] 0.82) and 3 ([kappa] = 0.83), which were rated as "Well" preserved. Procedure 4 yielded in a "Fair" preservation only ([kappa] = 0.71).

RBC counts were difficult to preserve properly. The day 0 agreement (between counts on day 0, am, and day 0, pm) for RBC counts was 81-88% in all procedures ([kappa] coefficients, 0.68-0.78), inferior to that of all other particles. Both FP and FN rates began increasing on days 1-3, depending on the procedure. The best preservation was achieved with procedure 2 (shown in detail in Fig. 1). At day 1, the [kappa] coefficient was 0.78 compared with 0.61 for refrigeration (procedure 1) or 0.66 without preservation (procedure 5; data available at Clinical Chemistry Online website). After 3 days, none of the procedures performed optimally ([kappa] range, 0.24-0.61; Table 2). The possible effect of low conductivity (low osmolality) on the lysis of RBCs was assessed in these urines (our correlation coefficient of conductivity to osmolality was r = 0.87 in another sample of 100 patient urines; unpublished data). With procedure 1, five originally RBC-positive specimens were scored as FN after 1 day of storage at 4 [degrees]C. The mean conductivity of these FN specimens was 12.9 mS/cm, compared with a mean conductivity of 13.7 mS/cm for the remaining 83 urines investigated (P = 0.20, Student t-test for two groups assuming unequal variances). Isoosmotic urine has a conductivity of ~10 mS/cm.

[FIGURE 1 OMITTED]

WBCs were well preserved in the measured adult urine specimens at 4 [degrees]C for 3 days (agreement, 91.3%; [kappa] = 0.86; Table 2). Similar results were obtained with procedures 2-4 at room temperature ([kappa] coefficients, 0.84, 0.80, and 0.80, respectively). The extent of deterioration without preservation (procedure 5) was also fairly low in this study ([kappa] = 0.67).

Casts were not a problem in preservation, as evaluated from the stability of counts after preservation (Table 2). Because of the low counts in the gray zone and small amounts of positive specimens (3-6 of 93 samples, depending on the procedure), [kappa] coefficients remained low because of random variation. Large ECs were fairly well preserved even without preservatives (procedure 5); no differences were seen between procedures 2 and 1. SRCs again showed low [kappa] coefficients because of the small number of positive specimens (7-8 of 93 cases were positive for SRC), despite an agreement >90% in all cases (Table 2).

VISUAL MICROSCOPY

For visual microscopy, 10 specimens were not followed because of the original low total count (nothing to be seen). At 3 days after collection, procedure 2 (BD solution) preserved most of the sediment particles better or at least as well as the other procedures when classification A or B was considered acceptable (Table 3). Procedures 1 and 3 were the next best options. Differences in the percentages did not reach statistical significance, mostly because of the low total number of specimens positive for each particle (8-62 cases). Only renal tubular cells were preserved better with procedure 3 (10 mL/L formalin-0.15 mol/L NaCl solution) than with procedure 2. No increase from the background value of 63% (procedure 5) was seen in preservation of hyaline casts by any procedure.

Discussion

Preservation of urine is a prerequisite for regional organization of specimen workflow and improvement in the quality (accuracy) of results. We wanted to examine the possibility of stabilizing urine for automated flow cytometry, which otherwise fulfills the clinical need for basic particle analysis of urine (1, 2, 12). Preliminary results from successful preservation with 5 mL/L formaldehyde (final concentration) for automated analysis were available (16).

Mean bacterial (BACT) counts were mostly reliable up to 3 days with procedures 1-3 (Tables 1 and 2). Procedure 1 (refrigeration) is the current reference preservation procedure for bacterial culture but is not always applicable in routine applications (8). Formaldehyde preservation (procedure 3) does not allow any culturing. With procedure 2 (BD C&S tube), a 3.4% FP rate was observed on day 3 compared with 28% with no preservation (procedure 5). Because of this slight FP rate with procedure 2 at day 3, specimen preservation should not be extended to 3 days (the company currently claims that the product is useful for bacterial culture for 2 days, which was not examined in this study).

Our difficulty in preserving urine RBCs in general was somewhat surprising (Tables 1 and 2). RBCs were also less well preserved than granulocytes in visual microscopy (Table 3), indicating the fragility of these cells in urine. Lysis of RBCs was not related to low osmolality in the urines, as judged from the conductivity of the five specimens that lost their RBCs during storage. Preservation of RBCs succeeded slightly better with procedure 2 than with the other procedures for 1-3 days (Fig. 1 and Table 2). The reason for the stabilizing effect of the BD additives remains uncertain. Procedure 3 (10 mL/L formaldehyde0.15 mol/L NaCl solution) reduced RBC counts in our hands and created FN results as early as day 1 (Table 1). Our results agree with those of Hawkins et al. (16), who showed that RBCs shrink in formaldehyde solutions.

WBC preservation succeeded better than expected in our experiments (Tables 1 and 2). In visual examination, WBCs were identified in 19% of specimens without preservation (procedure 5), but 79% and 82% of specimens could be classified after storage with procedures 1 and 2 for 3 days, respectively (Table 3). These results may reflect our selection of morning urine from adult patients. Preservation of WBCs is inferior in pediatric (random) specimens (5).

Most urine casts were stable with procedures 1 and 2, but also in specimens without preservatives (Table 2). Procedures 3 and 4 created artifactual increases in the automated counts (Table 1). In visual microscopy, most of the hyaline casts were preserved only partially despite the various preservation procedures (Table 3). Here, repeated automated analysis may have been more gentle than repeated mixing of vials for sediment microscopy on day 3. Casts are known to disintegrate in urine with alkaline pH (17), whereas our specimens were mostly acidic morning urines.

Large (ECs) and small (SRCs) epithelial cells were fairly well preserved in automated counting (Table 2); only procedure 3 (formaldehyde) created artifactually high counts on day 3 (Table 1). Detailed visual microscopy of ECs requires proper fixation if specimens are to be studied after 3 days (Table 3).

In conclusion, we have shown that urine particles can be preserved for regional transportation at room temperature, usually lasting for a maximum of 1 day. Preservation over the weekend should be considered carefully because of the risk of artifactual bacterial growth. For some particles, such as RBCs, preservatives seem to work better than refrigeration. The best preservation results were obtained with a commercial mixture originally developed for preservation of specimens for bacterial culture (10, 11). Undoubtedly, related preservatives have already been used for particle analysis, particularly in countries where microbiology laboratories also count the cells in urine. It must be noted that pediatric specimens, which have low osmolality and, frequently, high pH, were not analyzed in this study. Additional patient populations should also be examined to complete our knowledge on procedures for urine preservation.

We are indebted to Dr. Wolfgang Hofgartner of BD Preanalytical Solutions for constructive comments on the manuscript and to Dr. Vladimir Mats of BD Corporate Statistics for help in the evaluation of the significance of the results. This study was supported financially by BD Preanalytical Solutions.

Received October 2, 2001; accepted March 8, 2002.

References

(1.) Ben-Ezra J, Bork L, McPherson RA. Evaluation of the Sysmex UF-100 automated urinalysis analyzer. Clin Chem 1998;44:92-5.

(2.) Kouri TT, Kahkonen U, Malminiemi K, Vuento R, Rowan RM. Evaluation of Sysmex UF-100 urine flow cytometer vs. chamber counting of supravitally stained specimens and conventional bacterial cultures. Am J Clin Pathol 1999;112:25-35.

(3.) Koivula T, Gronroos P, Gavert J, Icen A, Irjala K, Penttila I, et al. Basic urinalysis and urine culture: Finnish recommendations from the working group on clean midstream specimens. Scand J Clin Lab Invest 1990;50(Suppl 200):26-33.

(4.) National Committee for Clinical Laboratory Standards. Urinalysis and collection, transportation, and preservation of urine specimens; approved guideline. NCCLS Document GP16-A. Wayne, PA: NCCLS, 1995.

(5.) Kierkegaard H, Feldt-Rasmussen U, Hoerder M, Andersen HJ, Joergensen PJ. Falsely negative urinary leukocyte counts due to delayed examination. Scand J Clin Lab Invest 1980;40:259-61.

(6.) Schumann GB, Schumann JL, Marcussen N. Cytodiagnostic urinalysis of renal and lower urinary tract disorders. New York: Igaku-Shoin, 1995:13.

(7.) Anpalahan M, Birch D, Becker G. Chemical preservation of urine sediment for phase contrast microscopy examination. Nephron 1994;68:180-3.

(8.) Clarridge JE, Johnson JR, Pezzlo MT. Cumitech 2B. Laboratory diagnosis of urinary tract infections. Washington DC: American Society for Microbiology, 1998:6.

(9.) Porter IA, Brodie J. Boric acid preservation of urine samples. Br Med J 1969;ii:353-5.

(10.) Watson PG, Duerden BI. Laboratory assessment of physical and chemical methods of preserving urine specimens. J Clin Pathol 1977;30:532-6.

(11.) Lauer BA, Reller LB, Mirrett S. Evaluation of preservative fluid for urine collected for culture. J Clin Microbiol 1979;10:42-5.

(12.) Langlois MR, Delanghe JR, Steyaert SR, Everaert KC, de Buyzere ML. Automated flow cytometry compared with an automated dipstick reader for urinalysis. Clin Chem 1999;45:118-22.

(13.) Sternheimer R. A supravital cytodiagnostic stain for urinary sediments. JAMA 1975;231:826-32.

(14.) Fleiss JL. Statistical methods for rates and proportions. New York: Wiley & Sons, 1981.

(15.) Kouri T, Fogazzi G, Gant V, Hallander H, Hofmann W, Guder WG, eds. European urinalysis guidelines. European Confederation of Laboratory Medicine. Scand J Clin Lab Invest 2000;60(Suppl 231):1-96.

(16.) Hawkins T, Ross M, McLennan J, Gyory AZ. Urine microscopy: its clinical value. Sysmex J Int 1996;6:41-5.

(17.) Burton JR. Quantitation of casts in urine sediment. Ann Intern Med 1975;83:518-9.

[4] Nonstandard abbreviations: RBC, red blood cell; WBC, white blood cell; EC, epithelial cell; SRC, small round (epithelial and inflammatory) cell; FN, false negative; and FP, false positive.

Timo Kouri, [1]* Lotta Vuotari, [2] Simo Pohjavaara, [2] and Pekka Laippala [3]

[1] Oulu University Hospital, Laboratory Administration, PO Box 50, FIN-90029 OYS, Finland.

[2] Centre for Laboratory Medicine, Tampere University Hospital, PO Box 2000, FIN-33521 Tampere, Finland.

[3] Tampere School of Public Health, Tampere University, and Research Unit of Tampere University Hospital, PO Box 2000, FIN-33521 Tampere, Finland.

* Author for correspondence. Fax 358-8-315-5541; e-mail timo.kouri@ppshp.fi.
Table 1. Preservation of particles counted with UF-100 flow
cytometer. (a)

 UF-100 particle code

 BACT RBC WBC

Original median count (x[10.sup.6]/L) 169 9 19
Original mean count (x[10.sup.6]/L) 339 41 102
Original minimum (x[10.sup.6]/L) 19 1 1
Original maximum (x[10.sup.6]/L) 2992 789 1731
Significance of changes in repeated <0.001 <0.001 0.362
 counts (b)
Deviating preservation procedures (c) 4 and 5 3

 UF-100 particle code

 CAST EC SRC

Original median count (x[10.sup.6]/L) 0.3 7.5 0.8
Original mean count (x[10.sup.6]/L) 0.6 13.9 1.5
Original minimum (x[10.sup.6]/L) 0.0 0.3 0.0
Original maximum (x[10.sup.6]/L) 7.2 186.8 14.7
Significance of changes in repeated 0.023 0.001 0.908
 counts (b)
Deviating preservation procedures (c) 3 3

(a) Particles shown: BACT (bacteria, diameter [greater than or
equal to]2 [micro]m), RBCs, WBCs, CAST (casts), EC (large
epithelial cells), and SRCs.

(b) Interaction term (procedure x time point) from analysis of
variance of Box-Cox-transformed data.

(c) Obtained from Tukey simultaneous tests.

Table 2. Preservation of particles counted with UF-100 flow
cytometer as evaluated from change of category from positive
to negative or vice versa. (a)

 UF-100 code

Preservation procedure BACT RBC WBC

1: Refrigerated, no preservatives Well Poor Well
2: Urine C&S tube (BD) Well Fair Well
3: 10 mL/L formalin with 0.15 mol/L NaCl Well Poor Well
4: 80 mL/L ethanol+ 20 g/L polyethylene Fair Poor Well
 glycol
5: No preservatives,+ 20 [degrees]C Poor Poor Fair

 UF-100 code

Preservation procedure CAST EC SRC

1: Refrigerated, no preservatives Well Well Fair
2: Urine C&S tube (BD) Fair Well Fair
3: 10 mL/L formalin with 0.15 mol/L NaCl Poor Fair Well
4: 80 mL/L ethanol+ 20 g/L polyethylene Well Well Poor
 glycol
5: No preservatives,+ 20 [degrees]C Poor Well Poor

(a) Limiting values of [kappa] coefficients after 3 days of
storage used for grading: high-count particles (BACT, RBC,
and WBC), well, [greater than or equal to]0.8; fair,
[greater than or equal to]0.6; poor, <0.6; low-count
particles (CAST, SRC, and EC), well, [greater than or equal
to]0.7; fair, [greater than or equal to]0.5; poor, <0.5.

Table 3. Preservation of visual microscopy results after 3 days
with different procedures.

 Samples still readable after 3 days, %

 RBCs Squamous
Preservation procedure Granulocytes erythrocytes ECs

1: Refrigerated, no 79 65 92
 preservatives
2: Urine C&S tube (BD) 82 71 95
3: 10 mL/L formalin 81 63 92
 with 0.15 mol/L NaCl
4: 80 mL/L ethanol + 60 53 92
 20 g/L polyethylene
 glycol
5: No preservatives, + 19 37 89
 20 [degrees]C

No. of samples positive 62 49 37
for each finding at
arrival (total n = 96)

 Samples still readable after 3 days, %

 Transitional Renal tubular Hyaline
Preservation procedure ECs cells casts

1: Refrigerated, no 76 64 63
 preservatives
2: Urine C&S tube (BD) 84 55 63
3: 10 mL/L formalin 72 73 63
 with 0.15 mol/L NaCl
4: 80 mL/L ethanol + 64 45 63
 20 g/L polyethylene
 glycol
5: No preservatives, + 60 27 63
 20 [degrees]C

No. of samples positive 25 11 8
for each finding at
arrival (total n = 96)
COPYRIGHT 2002 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Hematology
Author:Kouri, Timo; Vuotari, Lotta; Pohjavaara, Simo; Laippala, Pekka
Publication:Clinical Chemistry
Date:Jun 1, 2002
Words:4043
Previous Article:Evaluation of the anticoagulants EDTA and citrate, theophylline, adenosine, and dipyridamole (CTAD) for assessing platelet activation on the ADVIA...
Next Article:Improved method for gas chromatographic--mass spectrometric analysis of 1-[sup.13]C-labeled long-chain fatty acids in plasma samples.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters