Manual versus computer-automated semen analyses. Part I. Comparison of counting chambers

FERTILITY AND STERILITY Vol 65, No I, January 1996 Copyright c 1996 American Society for Reproductive Medicine Printed on acid free paper in U S A Manual versus computer-automated semen analyses Part I Comparison of counting chambers Jane E Johnson, MT (ASCP)*t William R Boone, PhDt Dawn W Blackhurst, MS:!: Greenville Hospital System, Greenville, South Carolina Objective: To determine the accuracy and precision of counting chambers analyzed manually and with a computer-automated semen analyzer (CASA; Hamilton-Thorne Research, Beverly, MA) Design: Prospective study using comparative measurements of sperm concentration, motility and concentration of latex beads with three types of counting chambers: hemacytometers, 12 and MicroCell (Conception Technologies, Inc, La Jolla, CA) Chambers, and Malder (Sefi Medical Instruments, Haifa, Israel) Chambers Setting: A hospital-based Andrology laboratory Patients: Male partners of couples undergoing infertility evaluation Main Outcome Measures: Experiment I: measurements of sperm concentration were evaluated within hemacytometers; Experiment II: measurements of sperm concentration in the 35 to 50 X 10 6 /ml range and sperm motility were determined using counting chambers; Experiment III: accuracy of counting chambers was determined using a known concentration of beads Results: Experiment I: differences were demonstrated within two of eight hemacytometers (side 1 versus side 2); however, no significant effect of hemacytometer on variation in sperm concentration measurements was observed Experiment II: CASA-analyzed MicroCell Chambers demonstrated the best precision for sperm concentration (intraclass correlation coefficient = 093) and motility (intraclass correlation coefficient = 088) Experiment III: MicroCell Chambers most accurately determined known bead concentration (35 ± 5 X 10 6 /ml) whether analyzed on CASA (349 X 10 6 /ml) or manually (352 X 10 6 /ml) Conclusions: MicroCell Chambers proved to be accurate and precise for determining concentration and motility of semen specimens whether analyzed manually or with CASA Fertil Steril 1996;65:150-5 Key Words: Sperm, CASA, quality control, hemacytometer, Makler Chamber, MicroCell Chamber, counting chambers, beads, validation Results from semen analyses are used as indicators of male fertility The routine semen analysis includes measurements of volume, ph, white blood cell concentration, sperm agglutination, sperm motility, sperm concentration, and sperm morphology Received November 1, 1994; revised and accepted July 26, 1995 * Reprint requests: Jane E Johnson, MT (ASCP), Greenville Hospital System, Reproductive Endocrinology Associates, 890 West Faris Road, Suite 470, Box 2, Greenville, South Carolina 295 (FAX: 3-455-8492) t Reproductive Endocrinology Associates :j: Division of Medical Education and Research Traditionally, sperm motility values have been obtained through subjective visual assessment of a wet preparation, and a hemacytometer has provided sperm concentration values Despite reports that discrepancies exist between hemacytometers (1, 2), there have not been, until recently, other commercially available counting chambers designed for determination of sperm concentration The Makler Chamber (Sefi-Medical Instruments, Haifa, Israel) and the disposable Micro Cell Chamber (Conception Technologies, Inc, La Jolla, CA) provide new means of determining both sperm concentration and motility A second feature 150 Johnson et ai Comparison of sperm counting chambers Fertility and Sterility

of these chambers is their application to both manual and computer-automated semen analysis (CASA) This report details our validation studies of three different types of counting chambers Our purpose is to compare the accuracy and precision of several different counting chambers analyzed manually and with CASA Study Population MATERIALS AND METHODS The study population consisted of patients presenting to Reproductive Endocrinology Associates of the Greenville Hospital System for semen analysis All patients signed and received a copy of a consent form detailing the aim of the study and semen specimen handling procedures This study proposal was presented to and approved by the Institutional Review Committee of the Greenville Hospital System Experimental Protocol Upon receipt by the laboratory, semen specimens were placed into a 37 C incubator for up to 30 minutes to allow liquefaction to take place Specimens were vortexed gently and evaluated for volume and ph Aliquots of the specimens were evaluated for sperm agglutination, motility, concentration morphology, and concentration of white blood cells The remainder of each specimen was dedicated to this study Study specimens were analyzed as time permitted, generally 3 to 4 hours after collection This time delay resulted in a decrease in the average motility of specimens from that of their original analysis Parameter Settings for CASA The importance of reporting CASA parameter settings has been alluded to in a recent article (3) The parameter settings we used for semen analysis in Experiment II are listed below, as are the parameter settings used for evaluating latex beads in Experiment III Nine fields and a minimum of 0 sperm cells or beads were analyzed per specimen The standard parameter settings used with the Hamilton-Thorne Internal Visual Optical System (NOS; Hamilton-Thorne Research, Beverly, MA) were as follows: frames acquired: 30; frame rate: 30/ s; minimum contrast: 8; minimum size: 6; LOIRI size gates: 06 to 16; LOIRI intensity gates: 06 to 16; nonmotile head size: 10; nonmotile brightness (head intensity): ; medium path velocity (V AP) value: 25; low YAP value: 10; slow cells motile: yes; and threshold straightness (STR): The high-density parameter settings used with the IVOS were as follows: frames acquired: 7; frame rate: 30/s; minimum contrast: 8; minimum size: 6; LOIRI size gates: 06 to 16; LOIRI intensity gates: 06 to 16; nonmotile head size: 10; nonmotile brightness (head intensity): ; medium V AP value: 25; low V AP value: 10; slow cells motile: yes; and threshold straightness (STR): The parameter settings for analysis of latex beads on the NOS were as follows: frames acquired: 5; frame rate: /s; minimum contrast: 8; minimum size: 7; LOIRI size gates: 05 to 20; LOIRI intensity gates: 04 to 16; nonmotile head size: 8; nonmotile brightness (head intensity): 24; medium YAP value: 25; low V AP value: 10; slow cells motile: yes; and threshold straightness (STR): Statistical Methods All data were analyzed using SAS statistical software (SAS Institute, Inc, Cary, NC) Statistical tests used for each experiment are detailed in the methodology sections below Experiment I-Evaluation of Within-Hemacytometer Variation Methodology for the use of hemacytometers has been described in detail elsewhere (4) Two investigators participated in data collection Initially, four hemacytometers (Baxter Healthcare Corporation, McGaw Park, IL) were used in this experiment (A, B, C, and D) With prolonged use, two of the four hemacytometers began to show signs of wear, ie, erosion of the grid lines We became concerned that this could affect sperm concentration results and substituted four new hemacytometers (E, F, G, and H) for the four original hemacytometers In total, 366 semen specimens were evaluated on side 1 and side 2 of each hemacytometer (A, 72 specimens; B, 50 specimens; C, 72 specimens; D, 62 specimens; E, 30 specimens; F, 27 specimens; G, 30 specimens; and H, 25 specimens) Each specimen was assigned randomly to a single hemacytometer The specimens ranged in sperm concentration from 26 to 3355 X lo B /ml The mean difference between counts on the two sides of each hemacytometer was calculated and then paired t-tests were used to determine whether significant differences in counts of sperm concentration existed between side 1 and side 2 of each hemacytometer Analysis of variance procedures, account- Vol 65, No1, January 1996 Johnson et al Comparison of sperm counting chambers 151

ing for donor variation, were used to assess the effects of investigator and hemacytometer Experiment II-Comparison of Counting Chambers Researchers have reported that CASA instruments perform best within a sperm concentration range of > X 10B/mL (4) and <50 X 10B/mL (5) Based on these data, semen specimens with sperm concentrations greater than the upper range limit were diluted with seminal plasma to bring them within our chosen sperm concentration range of approximately 35 to 50 X 10B/mL Before each analysis, specimens were vorlexed gently, and appropriate aliquots of semen were removed from the sterile collection container, diluted, and placed on hemacytometers (the "gold standard" used for assessment of accuracy) for analysis of sperm concentration In addition, Makler Chamber and 12 and MicroCell Chambers were loaded using specific volumes of semen (42, 30, and 50 p,l, respectively), and then analyzed for sperm concentration and motility using a compound microscope and the CASA On a total of 10 specimens, this procedure was repeated three times per specimen by each of two investigators (two counts per hemacytometer; one manual and one CASA count for each Makler Chamber and 12 and MicroCell Chamber; three replicates per investigator; two investigators and 10 specimens for a total of 4 counts of sperm concentration and 3 counts of sperm motility) We used Pearson correlation coefficients as a measure of agreement between the two investigators and intraclass correlation coefficients as measures of precision (6); these coefficients were calculated from the results of an analysis of variance components procedure for a general linear model (PROC V ARCOMP in SAS) (7) Experiment III-Evaluation of Quality Control Beads D sing a single lot of a known concentration of beads (35 ::!:: 5 X 10B/mL) suspended in an aqueous solution, we evaluated three different counting chambers The latex beads (ACCD-BEADS; Hamilton-Thorne Research, Beverly, MA) were approximately the same size as human sperm heads The three counting chambers tested were the hemacytometer (four different counting chambers), the Makler Chamber (two different counting chambers), and the MicroCell Chamber ( different chambers) (Experiment II demonstrated no significant difference between the performances of the 12 and MicroCells; therefore, the MicroCell Chamber was chosen for use in this experiment) A total of aliquots of the bead suspension were counted on each of the three chamber types Beads on the hemacytometers were counted manually, on sides 1 and 2 of each hemacytometer, with the aid of a compound microscope, while a single count of the beads was performed, manually and on the CASA, using the Makler Chambers and the MicroCell Chambers Single counts were performed to minimize error due to specimen evaporation One-sample t-tests were used to assess differences between the known bead concentration and the bead concentration determined by each method and counting chamber RESULTS Experiment I-Evaluation of Within-Hemacytometer Variation Data demonstrated significant differences within (side 1 versus side 2) two of the eight hemacytometers tested (C and H) and approached significance with a third hemacytometer (A; P = < 001,003, and 008, respectively), but no overall effect of different hemacytometers on variation in sperm concentration measurements (P = 049) was observed We also found no significant technician effect on these data (P = 092) Experiment II-Comparison of Counting Chambers Figure 1 presents the results of two investigators' measurements of sperm concentration plotted against each other and the corresponding Pearson correlation coefficient (r) for the investigators and intraclass correlation coefficient for the different chambers tested The Pearson correlation coefficients indicated that the investigators demonstratl;ld the best agreement using the CASA (r = 085 to 09"3) compared with the manual methods (r = 002 to 082) regardless of the counting chamber Between counting chambers, the MicroCells outperformed the hemacytometer and the Makler Chambers in both CASA and manual analyses The MicroCell Chamber analyzed on the CASA demonstrated the best precision (intraclass correlation coefficient = 093), followed by the CASA-analyzed 12 Micro Cell Chamber (intraclass correlation coefficient = 091) The poorest performer was the Makler Chamber whether analyzed using the CASA (intraclass correlation coefficient = 0) or manually (intraclass correlation coefficient = 016) For the MicroCell Chambers, the CASA method for de- 152 Johnson et a1 Comparison of sperm counting chambers Fertility and Sterility

A Hemacytometer 100 1=059 ICC 055, ', 40 ", -'i l 40 100 C Microcell, CASA 100 1=093 ICC=093 40 ' l-:i 40 100 E 12 Microcell, manual 100 1=082 ICC=075 40,, ~: 40 100 G Makler, manual 100 1=002 ICC=015, 40 ~ :::-,- '-----"-----'--'----'---" 40 100 B 12 Microcell, CASA 100 1=090 ICC=091 40 i, ': ', toe 40 100 D Makler, CASA 40 ", 100 1=085 ICC0,, 40 100 F Microcell, manual 100 1=078 ICC=O56 40 1=',' : - 40 100 Figure 1 Distribution of sperm concentration results (x 10 6 /ml) from the triplicate analysis of 10 semen specimens, manually and by CASA, by two investigators (Investigator 1 results on the x axes versus Investigator 2 results on the y- axes) (n = 30 per investigator) using four different counting chambers r, Pearson correlation coefficient; ICC, intraclass correlation coefficient) (Experiment II) termining sperm concentration ( MicroCell intraclass correlation coefficient = 093, 12 MicroCell intraclass correlation coefficient = 091) proved to be superior to the manual method ( MicroCell intraclass correlation coefficient = 066, 12 MicroCell intraclass correlation coefficient = 076) Table 1 contains descriptive statistics of sperm concentration measurements for the seven different methods of analysis Traditionally, the hemacytometer has been used as the "gold standard" for measurements of sperm concentration; therefore, we used it as the standard for comparison in calculating the 95% confidence intervals The only counting chamber that differed significantly from the hemacytometer was the Makler Chamber analyzed on the CASA (mean ::':: SD = 417 ::':: 97 versus 514 ::':: 193 X 10 6 /ml) The Makler Chamber values were consistently higher than the hemacytometer values The analysis of each study specimen took approximately 25 hours However, a specimen's loss ofmotility over time was not significantly different among the different counting chambers The highest percent loss of motility over time was demonstrated by the manually analyzed Makler Chamber (58%/min), and the lowest percent loss of motility over time was demonstrated by the manually analyzed Micro Cell Chamber (14%/min) Data in Table 2 demonstrate that the 12 and MicroCell Chambers performed similarly for motility regardless of whether the analyses were performed on the CASA (intraclass correlation coefficient = 085 and 088, respectively) or manually (intraclass correlation coefficient = 084 and 0, respectively) Because of the Makler Chamber's ability to provide more consistent sperm distribution than a wet preparation of semen, we had, before the completion of this study, routinely used it in our laboratory for manual motility determinations Therefore, we selected the Makler Chamber as the standard for comparison for motility determinations (Table 2) Only the MicroCell Chamber analyzed on the CASA provided motility results similar to those of the Makler Chamber ( MicroCell mean motility Table 1 Computer-Automated Semen Analyzer and Manual Measurements of Sperm Concentration Performed on Four Different Counting Chambers (Experiment 11)* Counting method No of samples liemacytometer 1 12 MicroCell, CASA MicroCell, CASA Makler, CASA 12 MicroCell, manual MicroCell, manual Makler, manual Sperm concentration xlo'/ml 417 ± 97 (245 to 690) 404 ± 161 (5 to 853) 368 ± 144 (21 7 to 2) 514 ± 193 (303 to 1130) 406 ± 129 (240 to 853) 380 ± 91 (244 to 692) 468 ± 130 (255 to 845) Confidence intervals -89 to 46 122 to 14 09 to 144t -85 to 50-107 to 28 * Values are means ± SD with ranges in parentheses t Significantly different from the hemacytometer -22 to 113 Vol 65, No1, January 1996 Johnson et ai Comparison of sperm counting chambers 153

Table 2 Computer-Automated-Semen Analyzer (CASA) and Manual Measurements of Sperm Motility Performed on Three Different Counting Chambers (Experiment 11)* Counting method No of samples Sperm motility 95% confidence interval Intraclass correlation coefficient Makler, manual 12 MicroCell, CASA MicroCell, CASA Makler, CASA 12 MicroCell, manual Microcell, manual % 320 ::t: 159 (21 to 593) 240 ::t: 138 (20 to 540) 271 ::t: 146 (30 to 610) 240 ::t: 133 (40 to 510) 389 ::t: 148 (88 to 630) 408 ::t: 139 (88 to 743) -146 to -13t -116 to 17-146 to -l4t 03 to 135t 21 to 154t 042 085 088 068 084 0 * Values are means ::t: SD with ranges in parentheses t Significantly different from the manually analyzed Makler Chamber = 271% ± 146%, Makler Chamber mean motility = 320% ± 159%) The MicroCell Chamber provided consistently higher percent motility values when analyzed manually (408% ± 139%) Compared with the Makler Chamber, the 12 MicroCell Chamber consistently provided a lower percent motility when used with the CASA (240% ± 138%) and consistently higher percent motility when analyzed manually (389% ± 148%) The Makler Chamber, itself, repeatedly provided lower motility counts when analyzed using the CASA (240% ± 133%) Experiment III-Evaluation of Quality Control Beads The concentration of the latex beads was determined using hemacytometers, Makler Chambers, and MicroCell Chambers (Fig 2) As in Experiment II, the CASA-analyzed MicroCell Chamber proved to be the most accurate (mean bead concentration = 349 ± 19 X 10 6 /ml) followed by the manually analyzed MicroCell Chamber (mean bead concentration = 352 ± 42 X 10 6 /ml) The hemacytometer underestimated the average bead concentration (mean = 333 ± 51 X 10 6 /ml) but was not significantly different from the known concentration of 35 ± 5 X 10 6 /ml However, the Makler Chamber significantly overestimated the concentration of the beads when they were counted either manually (mean = 475 ± 110 X 10 6 /ml) or on the CASA (mean = 492 ± 119 X 10 6 /ml; both P < 001) DISCUSSION The hemacytometer may no longer be the best choice as the "gold standard" determinant of sperm concentration It has been known for more than 50 years that considerable variation in estimates of cell concentration exist when a hemacytometer is used The sources of this variation include differences be- tween hemacytometers, dilution errors, and variations between technicians Berkson and co-workers (1) demonstrated an error of 46% between hemacytometers Our data from Experiment I suggest that variation also exists within hemacytometers Of the eight hemacytometers evaluated, two demonstrated significant differences between their two sides The Berkson study (1) also reported that an additional 47% error was associated with the dilution of specimens (pipetting error) Similar findings (43% chamber and pipetting error combined) were reported by Biggs and MacMillan (2) Biggs and MacMillan (2) also stated that independent measurements of cell concentration, even when performed by skilled technicians, show poor agreement Berkson and col- r------------------------~ 75 70 65 475;1; 110 492;1; 119 333;1;51 I 352;1;42 : I 349;1; 19 40 ---~---------~---~--------- I: I 35~--~~---+---------~--------~ 30 25 L----L----~--~-----L----L---~ Hemacy- Makler, 2Ou-CeIl, Makler, 2Ou-Cell tometer manual manual CASA CASA Figure 2 Distribution of results from counting, manually and by CASA, a known concentration (35 [solid line] ::t: 5 [dashed lines] X 10 6 ml) of latex beads using various counting chambers (Experiment III) 154 Johnson et al Comparison of sperm counting chambers Fertility and Sterility

leagues (1) concluded that the guidelines for required agreement between successive counts of cell concentration were " meaninglessly stringent Differences considered as too large to be allowed will occur very frequently (66% to 85% of the time) if counting is made precisely and recorded faithfully" The hemacytometer cannot meet the quality control requirements for precision desired by today's laboratories This is not only due to Chamber error but also may be due to technician-associated error (dilution and Chamber-filling errors) This latter error is subject to debate as good intertechnician agreement has been reported by some scientists (4, 8-10), whereas poor agreement has been reported by others (11, 12) In Experiment II, results indicate that the MicroCell Chamber is more precise than either the hemacytometer or the Makler Chamber for determining sperm concentration and motility This in part may be due to the fact that, unlike the other two chambers, the MicroCell Chamber fills by capillary action and cannot be overfilled, thus providing a more constant depth of semen in the chamber The greater precision ofthe MicroCell Chamber, as compared with the Makler Chamber, for determining sperm motility is in contrast to earlier published data that demonstrated no difference in the percent motility of semen specimens between the chambers (13) In Experiment III, the MicroCell is the most accurate and precise chamber for counting beads Again, because accuracy, like precision, is vital to quality control, the MicroCell Chamber appears to be the counting chamber of choice Based on the results of Experiment III, the rank of the various counting chambers for accuracy and precision is [1] the MicroCell Chamber analyzed on the CASA, [2] the MicroCell Chamber analyzed manually, [3] the hemacytometer, [4] the Makler Chamber analyzed manually, and [5] the Makler Chamber analyzed on the CASA Similar results, generated from more limited data, have been reported by others (11) We conclude that there is considerable variation within hemacytometers (side 1 versus side 2); the MicroCell Chamber should replace the hemacytometer as the gold standard for measurements of sperm concentration; and the Makler Chamber is a poor choice for determining sperm concentration The use of the MicroCell Chamber for determining sperm concentration and motility has been implemented already in our laboratory Acknowledgment The MicroCell Chambers were provided by Conception Technologies, Inc, La Jolla, California REFERENCES 1 Berkson J, Magath TB, Hurn M The error of estimate of the blood cell count as made with the hemacytometer Am J Physiol 1940; 128:309-23 2 Biggs R, MacMillan RI The errors of some haematological methods as they are used in a routine laboratory J Clin Pathol 1948; 1:269-87 3 Davis RO, Katz DF Standardization and comparability of CASA instruments J AndroI1992;13:81-6 4 Johnson JE, Boone WR, Shapiro SS Determination of the precision of an automated semen analyzer Lab Med 1990; 21:33-8 5 Mortimer D, Goel N, Shu MA Evaluation of the CellSoft automated semen analysis system in a routine laboratory setting Fertil Steril 1988;50:9-8 6 Winer BJ, Brown DR, Michels KM Statistical principles in experimental design 3rd ed New York: McGraw-Hill, 1991 7 SAS Institute Inc SAS/STAT user's guide, release 603 edition Cary (NC): SAS Institute Inc, 1988 8 Cooper TG, Neuwinger J, Bahrs S, Nieschlag E Internal quality control of semen analysis Fertil Steril 1992;58: 172-8 9 Dunphy BC, Kay R, Barratt CLR, Cooke ID Quality control during the conventional analysis of semen, an essential exercise J Androl 1989; 10:378-85 10 Mortimer D, Shu MA, Tan R Standardization and quality control of sperm concentration and sperm motility counts in semen analysis Hum Reprod 1986; 1:299-303 11 Jequier AM, Ukombe EB Errors inherent in the performance of a routine semen analysis Br J Urol 1983;55:434-6 12 Freund M, Carol B Factors affecting haemocytometer counts of sperm concentration in human semen J Reprod Fertil 1964; 8:149-55 13 Ginsburg KA, Armant DR The influence of chamber characteristics on the reliability of sperm concentration and movement measurements obtained by manual and videomicrographic analysis Fertil Steril 1990;53:882-7 Vol 65, No1, January 1996 Johnson et al Comparison of sperm counting chambers 155