- Email: [email protected]
Evaluation of the CellTrak Computer-Assisted Sperm Analysis System* in comparison to the Cellsoft System† to measure human sperm hyperactivation
FERTILITY AND STERILITY Copyright
©
1995 American Society for Reproductive Medidne
Vol. 64, No.2, August 1995
Printed on acid-free paper in U. S. A.
Evaluation of the CeliTrak Computer-Assisted Sperm Analysis System* in comparison to the Cellsoft Systemt to measure human sperm hyperactivation Evan H. Hurowitz Andrew Leung Christina Wang, M.D.:j: Division of Endocrinology, Department of Medicine, Harbor-University of California Los Angeles (UCLA) Medical Center, Torrance, California
Objective: To measure and compare sperm hyperactivation (HA) and associated motility parameters using the CellTrak (Motion Analysis Corporation, Santa Rosa, CAl and Cellsoft (Cryo Resources, New York, NY) Computer-Assisted Sperm Analysis (CASA) Systems. The accuracy of the CellTrak system also was assessed by repeated measurements of the same sample. Setting: A university-affiliated tertiary care center. Patients: Normal healthy volunteers. Main Outcome Measures: We assessed sperm motion parameters and HA. Results: The measurements of percent spermatozoa with star and/or transitional pattern of HA showed high within-sample variations, especially in samples showing low levels of percent spermatozoa with HA. This was due to the comparatively low numbers of hyperactivated spermatozoa that were present and counted in a sample. Excellent agreement between the two systems was observed for all parameters except for measurements of curvilinear velocity, percent spermatozoa with star-spin hyperactivity, and percent spermatozoa with transitional hyperactivity. The difference might be due to CellTrak's higher level of accuracy in determining velocity. Conclusions: The CellTrak system effectively measures human sperm HA provided adequate numbers of motile spermatozoa are analyzed. Although it suffers from the same limitations common to all CASA systems, it is more accurate than the Cellsoft system. Fertil Steril 1995; 64:427 -32 Key Words: Human spermatozoa, hyperactivated motility, computer-assisted sperm analysis
For mammalian sperm to fertilize an oocyte, they must undergo a final maturation process in the female reproductive tract. This process, known as capacitation (1, 2), includes a change in sperm motility characterized by low forward progression and high flagellar amplitude known as sperm hyperactivation (HA) (3). Although the relationship between the development of hyperactivated motility and the other
Received October 24, 1994; revised and accepted March 2, 1995. * CellTrak Computer-Assisted Sperm Analysis System, Motion Analysis Corporation, Santa Rosa, California. t Cellsoft Computer-Assisted Sperm Analysis System, Cryo Resources, New York, New York. :j: Reprint requests: Christina Wang, M.D., Clinical Study Center, Harbor-UCLA Medical Center, Torrance, California 90509 (FAX: 310-533-6972). Vol. 64, No.2, August 1995
stages of capacitation are not defined clearly, the lack of HA may relate with the inability of sperm to bind to the zona pellucida and to fertilize an oocyte in vivo. To understand completely the process of capacitation, it was necessary to develop methods of measuring HA in human spermatozoa. Unlike other qualitative measurements, such as sperm morphology, HA involves the motion of spermatozoa and the changes in that motion, making it very difficult and extremely time consuming to measure manually (4, 5). Current technology has combined videomicroscopy and computer-assisted digital image analysis to measure HA (6-9). In this study we compared two computer-assisted sperm analysis (CASA) systems, the newer CellTrak system and the more established Cellsoft system, to assess the reliability of the CellTrak system to assess HA. Hurowitz et a1. Human sperm hyperactivation
427
MATERIALS AND METHODS Sample Preparation
Semen samples were obtained from healthy normal male volunteers after ~48 hours sexual abstinence. Semen was collected by masturbation into sterile wide-mouth plastic containers provided by the laboratory. Samples were allowed to liquefy for 30 minutes at room temperature before processing. Semen was layered onto discontinuous Percoll gradients of80% and 55% (Pharmacia, Uppsala, Sweden), centrifuged, and the final pellet was washed in an N -[2-hydroxyethyl]piperazine-1-[2ethanesulfonic acid]-buffered Ham's F-10 medium with 1% bovine serum albumin (GIBCO, Grand Island, NY). The washed spermatozoa then were suspended at a concentration of 10 to 20 X 1Q6/mL in Ham's F-10 with 10% heat-inactivated pooled human serum obtained from preovulatory women undergoing IVF (capacitation medium) (9). Forty-two sperm preparations were assessed in this study. Sperm Motion Analysis and Hyperactivated Motility
Sperm motion analysis and assessment of HA were performed on spermatozoa incubated in capacitation media after 10 minutes, 1, 5, and 24 hours. Eight microliters of the sperm suspension was placed into 20-J-lm Microcell slides (Fertility Technologies, Natick, MA). All sperm preparations were analyzed on a heated microscope stage (37°C) and recorded by video for ~ 10 minutes. The Cellsoft CASA System (Version 3.51c; Cryo Resources, New York, NY) equipped with a HA module was used for analysis of the videotapes with setup parameters as described previously (7 -9). The videotapes of the sperm preparations were then reanalyzed using the CellTrak CASA System (Version 4.25 Motion Analysis VP 110; Motion Analysis Corporation, Santa Rosa, CA). The following parameters were computed or calculated: curvilinear velocity (VCL, J-lm/s) , straight line velocity (VSL, J-lm/s) , linearity (LIN = VSLNCL X 100), maximum amplitude of lateral head displacement (ALHmru" J-lm), mean amplitude of lateral head displacement (ALHmeam J-lm), and head beat cross frequency (Hz). In addition to the parameters described above, the derived parameters were calculated for each cell: DANCE MEAN = ALHmean X VCLNSL (J-lm). Hyperactivated sperm cells were classified as transitional if the VCL ~ 80 J-lm/s, LIN > 19, and :s; 34 J-lm/s, and DANCEMEAN ~ 17 J-lm, and as star-spin HA if the VCL ~ 80 J-lm/s, LIN :s; 19 J-lm/s, and DANCEMEAN ~ 17 J-lm according to the criteria described previously (6). Total HA is the sum 428
Hurowitz et al. Human sperm hyperactivation
of transitional HA and star-spin HA. Each individual cell track was analyzed ~23 times in 1 second, with a frame rate of 30 frames/so The data from each individual cell track then were recorded and printed. Two hundred to 800 motile cells were analyzed for each sample. Hyperactivated sperm cells were classified and HA data were presented under the criteria as described in this paragraph. Statistical Analyses
Mean, SD, and coefficients of variation (CVs) were computed for several measurements of the same sample by the CellTrak system. The results of all 42 samples for each parameter from each system were compared for agreement using one-factor analysis of variance, linear regression, and the method described by Bland and Altman (10). RESULTS Reproducibility of HA Assessment by CellTrak System
To determine whether the results from the CellTrak system were reproducible, four different samples with extremely low, low, average, and high percentages of hyperactivated spermatozoa (previously determined by the CellSoft system) were analyzed repeatedly by the CellTrak system. The results are summarized in Table 1. The results showed that the sperm motion parameters (percent motility, VCL, VSL, LIN, and ALH) were extremely reproducible, showing low CVs in all parameters. In the HA measurements, the repeated measurements from the high level samples show lower CVs compared with the low level samples. However, even with a mean total HA of 18.8% in sample 4, the CV remained high at 15%. The high CV was due to the small numbers of spermatozoa with HA in each measurement. Counting even higher numbers of sperm might decrease the severity of this variation. The repeated measurements of sample 4 in which 800 motile cells were counted showed further reduction in the CV compared with the measurements where 400 cells were counted for all parameters except one, the percent spermatozoa with transitional HA. The CV for total HA was reduced to <10%. At low HA levels, the system rarely gave a change more than ± 1 cell from the mean number of spermatozoa showing HA; whereas at high levels the computer may yield values up to ±6 cells from the mean. Thus the CellTrak system was much more precise at low levels, but might be less accurate, because ±6 is only 33% of 18 whereas ± 1 is 100% of ± 1 cells. In comparing the highest combination of accuracy and Fertility and Sterility
Table 1 Reproducility of Sperm Motility and HA Parameters as Assessed by the CellTrak CASA System*
Sample It (n
=
13):1:
Sample 2t (n
=
11)
Sample 3t (n
=
8)
Sample 4t (n
=
11)
Sample 4§ (n
=
10)
Percent motility
VCL
VSL
I'm/s
I'm/s
74 ::!: 1.05 (1.43) 80 ::!: 1.64 (2.05) 93 ::!: 1.04 (1.25) 88::!: 1.13 (1.28) 88 ::!: 0.67 (0.76)
70.8 ::!: 2.17 (3.06) 82.6 ::!: 2.45 (2.97) 97.8::!: 3.06 (3.13) 114.8 ::!: 4.88 (4.25) 113.1 ::!: 1.96 (1.73)
48.7 ::!: 2.01 (4.14) 48.1 ::!: 2.20 (4.56) 55 ::!: 2.14 (3.90) 54 ::!: 1.68 (3.10) 53 ::!: 0.94 (1.79)
LIN
ALH I'm
%
%
%
67.8 ::!: 2.43 (3.59) 57.9 ::!: 1.20 (2.08) 60.6 ::!: 2.05 (3.39) 52.1 ::!: 1.51 (2.91) 51.8 ::!: 0.94 (1.82)
3.8 ::!: 0.22 (5.89) 5 ::!: 0.18 (3.65) 5.5 ::!: 0.24 (4.45) 6.8 ::!: 0.28 (4.05) 6.8 ::!: 0.19 (2.77)
0.4 ::!: 0.38 (96.20) 1.6 ::!: 0.88 (55.20) 3.5 ::!: 1.63 (46.73) 7.6::!: 1.63 (21.44) 6.6 ::!: 2.33 (35.48)
0.19 ::!: 0.61 (67.89) 3.9 ::!: 1.29 (32.90) 7.6 ::!: 1.89 (24.83) 11.2 ::!: 1.57 (13.96) 12.3 ::!: 1.49 (12.19)
1.3 ::!: 0.57 (43.63) 5.5 ::!: 1.53 (27.74) 11.1 ::!: 2.10 (18.86) 18.8 ::!: 2.84 (15.09) 18.8 ::!: 1.83 (9.74)
* Values are means::!: SD with the CV, expressed as a percentage in parentheses. t For each parameter, ",,400 motile spermatozoa were counted.
Star-spin HA
Total HA
:I: Number of repeated assessments on the sample. § For each parameter, at ",,800 motile spermatozoa were counted.
preCISIOn samples with HA, between 8% and 10% tend to have the highest combination of accuracy and precision.
DISCUSSION
Computer-assisted sperm analysis systems provide a standardized and unbiased system of measuring and studying human sperm hyperactivated motility and they greatly improve the speed of a slow and tedious manual process (4-9) especially useful in clinical investigations_ Computer and image analysis technology is advancing and CASA technology still is being improved. In our study, even though all of the parameters associated with HA showed small variations when the same sample was analyzed repeatedly, the measurements of transitional HA, star-spin HA, and total HA showed high CVs. This was because low numbers of hyper activated spermatozoa were being counted during each analysis, despite the fact that 400 motile spermatozoa were analyzed for each sample. No sample in this experiment had >20% spermatozoa with HA, which meant that the CellTrak system analyzed <80 hyperactivated spermatozoa per sample. A few more or less spermatozoa counted during each assessment would cause large changes in percent of spermatozoa classified as exhibiting HA, especially in samples with lower levels of HA. Given this limitation, we suggest that as CASA systems become faster and more automated it would be prudent to count ~400 motile
Comparison of HA Assessment of CellTrak With Cellsoft CASA Systems
The values for percent motility, VCL, VSL, LIN, transitional HA, star-spin HA, and total HA for the same 42 samples were measured by each system and compared. Table 2 shows that mean ± SD of the 42 measurements for each parameter by each system as well as linear regression equations describing the relationships of the motion parameters assessed by the two systems. The averages of and the differences between the two assessments of each sample (Cellsoft versus CellTrak) were calculated and plotted for each of the 42 samples (Fig_ 1) to demonstrate the agreement of the results as measured by the two systems_ Good agreement was demonstrated for most of the motion analysis parameters_ The results showed the VSL and VCL measured by the CellTrak system was significantly higher than those measured by the Cellsoft system. The percent of spermatozoa with transitional HA was significantly lower and the percent of spermatozoa with star-spin HA was significantly higher when assessed by the CellTrak when compared with the Cellsoft system_
Table 2
Transitional HA
Comparison of Motion and HA Parameters Measured by the CellTrak and Cellsoft Systems in 42 Semen Preparations Parameter
CellTrak* 81.4 91.4 51.6 58.5 2.6 4.9 7.5
Percent motility VCL (p,m/s) VSL (p,m/s) LIN Transitional HA (%) Star-spin HA (%) Total HA (%)
±
4.41
± 12.1
± ± ± ±
±
7.43 4.91 1.7 2.7 3.7
Cellsoft* 79.4 77.8 46.9 59.7 4.3 2.9 7.2
± 5.3 ± 15.lt ± 1l.6t ±
± ± ±
5.1 2.9t 2.5t 4.1
Equation
r
0.82 0.89 0.85 0.73 0.50 0.36 0.77
CellTrak CellTrak CellTrak CellTrak CellTrak CellTrak CellTrak
= = = = = = =
0.69 0.71 0.54 0.70 0.30 0.38 0.70
x x x x
Cellsoft Cellsoft Cellsoft Cellsoft x Cellsoft x Cellsoft x Cellsoft
+ + + + + + +
26.8 36.1 26.2 16.9 1.31 3.80 2.43
t p < 0.01 when compared with CellTrak.
* Values are means ± SD. Vol. 64, No.2, August 1995
Hurowitz et al.
,
....... " .... . ''''
_ ...
Human sperm hyperactiuation
429
A. % Motility
~
- ==...
'i~
u
u
III
..
U C
..
! u
~ c
B.VCl
4.0 ..,.,..............................~..................-.. ·"'o~.. ·-O~ . ·., 2.0 0.0 -2.0 +-"'O:--:..--"eF-1,-_r-~:F:-:---"---j -4.0 -6.0 -8.0 ·······················. ·o. :250 -10.0 o -12.0 +---_---+----+---~ 85.0 90.0 80.0 70.0 75.0
=
0.0
0 ~
+2S0
Ii:;'" -10.0
u u -II
..
-20.0
~U
-30.0
III
u:!: c: III
~ is
Mean
-2S0
-40.0 60.0
80.0
Average
C.VSL 1~:~
.
-.~--~.-2-s-D'
_~:~ +---=o:.......=(X)Ox-__O__B_~_"G---"O:....:r.Mr-ea=n., -10.0 -15.0
-20.0
8&p
~ ................. r;.
30.0
00
u
- 2S0
......... ..... .................. .
40.0
120.0
D. LIN
"--.-.-..-.-.-.-..-.-..-.-..-.. -. . o
100.0
Average
50.0
60.0
15.0
= 0 ~
10.0
"ii~
o u -II
5.0
u:!: c: III
0.0
III
..
~o
~ is
-10.0 45.0
70.0
00
-5.0 50.0
55.0
Average
60.0
65.0
70.0
Average
E. Transitional Hyperactivity
..
! '""I'ii = .!! 0
'ii
. ..
!:!. u
c
~
it:
i5
F. Star-Spin Hyperactivity 12 10 8 6 4 2 0 -2 -4
o ················;;0
+ 250
······00
u
l!
~
COco
Gj
o
o
o ~(:) .. 8. o
:i'
Mean
~ ;:: 0 ~
- 250
Gj
2
4
6
10
8
2III
u
c:
Average
~
~ is
G. Total Hyperactivity :i' u
l!
~
Gj
~ ;:: 0 ~
Gj
2III
u
c: ~
~ is
6 4 2 0 -2 -4 -6 -8 -10
0
Mean
0 0
-2S0
0 0
5
10
15
Mean
0 0
0 5
10
15
20
Average
430
0
0 0
-2S0
0 0
+2S0
0
0 0
0
+2S0
00 0
0
0 ··········0
Average
"e O " 0
6 4 2 0 -2 -4 -6 -8 -10
Hurowitz et al. Human sperm hyperactiuation
Figure 1 Averages ofthe measurements obtained by the Cell Trak and Cellsoft systems plotted against the differences between them for (Al percent motility, (Bl VCL, (e) VSL, (D) LIN, (El transitional HA, (F) star-spin HA, and (G) total HA. The solid line indicates the mean value of the differences and the dashed lines indicate :!:2 SD from the mean (10).
Fertility and Sterility
spermatozoa to minimize sampling error and the effects of small numbers of spermatozoa with HA on the computer system's analysis of human sperm HA. In comparing the two systems, all of the sperm motion parameters agreed very well except for VCL and the three measurements of HA. In the case of VCL, the CellTrak system consistently measures higher VCL (mean difference 13.7 J-Lrnls) than the Cellsoft system. This may not be an undesirable difference because there is an extremely high correlation between the two measurements of VCL. It is in fact understandable because the CellTrak system uses a more accurate algorithm to track sperm trajectories than the Cellsoft system, especially at higher velocities. The percent cells with transitional HA assessed by CellTrak was systematically lower than that assessed by Cellsoft. The converse was true with the assessment of percent cells with starspin HA. The total percent cells with HA assessed by the two systems was very similar. Because sperm cells move in a whipping and thrashing pattern during HA, a curvilinear pattern during progression as opposed to following a straight line in forward-progressing sperm cells, much of the actual distance the spermatozoa travel is not estimated when considering only how much distance between the two points the sperm traveled. Thus, the closer the CASA system's analyses follow the cell's true path, the higher and more accurate the value of velocity it will give (11). This effect becomes significant in measuring VCL, especially in sperm with a high VCL such those that are classified as star-spin. According to the classification of Robertson et al. (6), the VSL:VCL (LIN) ratio is critical in distinguishing between transitional and star-spin HA patterns. The higher VCL measured by CellTrak system at higher sperm velocities will lower the LIN and classify more spermatozoa as star-spin HA rather than as transitional HA. It should be noted the total HA was comparable as measured by the two systems with no systematic bias. Recognizing the limitations of the CellTrak system and of CASA systems in general (12, 13), we conclude that the technology does effectively provide a quick and unbiased measure of hyperactivity in a semen sample. The CellTrak system showed good reproducibility and it showed a high degree of agreement with the Cellsoft system with most parameters. Beyond limitations common to all CASA systems (12, 13), we found additional problems using CASA systems to measure sperm HA. These are the number of spermatozoa counted and the algorithm used to track sperm trajectories. The former is the reason for the very high variations in transitional HA and star-spin HA measurements. Given those Vol. 64, No.2, August 1995
variations, it is impressive that the measurements of total HA display such a high degree of agreement. Nevertheless, this limitation can be minimized by counting larger numbers of sperm during each sample analysis. The latter limitation introduces a tendency for the computer to underestimate velocity (both VCL and VSL). Thus, the measurements of VCL and VSL estimated by the CellTrak system are probably more accurate than those made by the Cellsoft system. As CASA technology improves and even more accurate approximations to the actual sperm trajectory are developed, such inaccuracies in velocity may decrease, allowing more accurate measurements of HA. The clinical usefulness of measurements of sperm HA has to be examined further in studies to assess the predictive values of these sperm parameters in both assisted and natural human fertilization settings.
Acknowledgments. Evan H. Hurowitz* was a summer high school student supported by the Harbor Collegium Fund. We thank John Greaves, Ph.D., Motion Analysis Corporation (Santa Rosa, California), for responding to our request to develop and allowing us to test the software for assessing HA on the CellTrak CASA system.
REFERENCES 1. Austin CR. Observations on the penetration ofthe sperm into the mammalian egg. Aust J Sci Res B 1951;4:581-96. 2. Austin CR. The 'capacitation' of the mammalian sperm. Nature 1952; 170:326. 3. Yamagamachi R. Mammalian fertilization. In: Knobil E, Neill JD, Ewing LL, Markert CL, Greenwald GS, PfaffDN, editors. The physiology of reproduction. New York: Raven Press, 1988; 131-88. 4. Burkman LJ. Discrimination between nonhyperactivated and classical hyperactivated motility patterns in human spermatozoa using computerized analysis. Fertil Steril 1991; 55:363-71. 5. Mortimer ST, Mortimer D. Kinematics of human spermatozoa incubated under capacitating conditions. J Androl 1990; 11:195-203. 6. Robertson L, WolfDP, Tash JS. Temporal changes in motility parameters related to acrosomal status: identification and characterization of populations of hyperactivated human sperm. BioI Reprod 1988;39:797-805. 7. Wang C, Leung A, Tsoi WL, Leung J, Ng V, Lee KF, et al. Evaluation of human sperm hyperactivated motility and its relationship with the zona-free hamster oocyte sperm penetration assay. J Androl 1991; 12:253-7. 8. Uhler ML, Leung A, Chan SYW, Wang C. Direct effects of
* Acknowledgment from one of the referees: "I find the fact that the lead author was a summer high school student particularly exciting and worthy of special note. We need to encourage our peers to follow your path. Congratulations!" Hurowitz et al. Human sperm hyperactiuation
431
progesterone and antiprogesterone on human sperm hyperactivated motility and acrosome reaction. Fertil Steril 1992; 58:1191-8. 9. Wang C, Lee GS, Leung A, Surrey ES, Chan SYW. Human sperm hyperactivation and acrosome reaction and their relationships to human in vitro fertilization. Fertil Steril 1993;59:1221-7. 10. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307 -10.
432
Hurowitz et al. Human sperm hyperactivation
11. Mortimer D, Serres C, Mortimer ST, Jouannet P. Influence of image sampling frequency on the perceived movement characteristics of progressively motile human spermatozoa. Gamete Res 1988;23:313-27. 12. Boyers SP, Davis RO, Katz DF. Automated semen analysis. Curr Probl Obstet Gynecol Fertil 1989; 12:165-200. 13. Davis RO. The promise and pitfalls of computer-aided sperm analysis. In: Overstreet JW, editor. Male infertility. Infertility and reproductive medicine clinics of North America. Philadelphia: W.B. Saunders Co., 1992:341-52.
Fertility and Sterility
©
1995 American Society for Reproductive Medidne
Vol. 64, No.2, August 1995
Printed on acid-free paper in U. S. A.
Evaluation of the CeliTrak Computer-Assisted Sperm Analysis System* in comparison to the Cellsoft Systemt to measure human sperm hyperactivation Evan H. Hurowitz Andrew Leung Christina Wang, M.D.:j: Division of Endocrinology, Department of Medicine, Harbor-University of California Los Angeles (UCLA) Medical Center, Torrance, California
Objective: To measure and compare sperm hyperactivation (HA) and associated motility parameters using the CellTrak (Motion Analysis Corporation, Santa Rosa, CAl and Cellsoft (Cryo Resources, New York, NY) Computer-Assisted Sperm Analysis (CASA) Systems. The accuracy of the CellTrak system also was assessed by repeated measurements of the same sample. Setting: A university-affiliated tertiary care center. Patients: Normal healthy volunteers. Main Outcome Measures: We assessed sperm motion parameters and HA. Results: The measurements of percent spermatozoa with star and/or transitional pattern of HA showed high within-sample variations, especially in samples showing low levels of percent spermatozoa with HA. This was due to the comparatively low numbers of hyperactivated spermatozoa that were present and counted in a sample. Excellent agreement between the two systems was observed for all parameters except for measurements of curvilinear velocity, percent spermatozoa with star-spin hyperactivity, and percent spermatozoa with transitional hyperactivity. The difference might be due to CellTrak's higher level of accuracy in determining velocity. Conclusions: The CellTrak system effectively measures human sperm HA provided adequate numbers of motile spermatozoa are analyzed. Although it suffers from the same limitations common to all CASA systems, it is more accurate than the Cellsoft system. Fertil Steril 1995; 64:427 -32 Key Words: Human spermatozoa, hyperactivated motility, computer-assisted sperm analysis
For mammalian sperm to fertilize an oocyte, they must undergo a final maturation process in the female reproductive tract. This process, known as capacitation (1, 2), includes a change in sperm motility characterized by low forward progression and high flagellar amplitude known as sperm hyperactivation (HA) (3). Although the relationship between the development of hyperactivated motility and the other
Received October 24, 1994; revised and accepted March 2, 1995. * CellTrak Computer-Assisted Sperm Analysis System, Motion Analysis Corporation, Santa Rosa, California. t Cellsoft Computer-Assisted Sperm Analysis System, Cryo Resources, New York, New York. :j: Reprint requests: Christina Wang, M.D., Clinical Study Center, Harbor-UCLA Medical Center, Torrance, California 90509 (FAX: 310-533-6972). Vol. 64, No.2, August 1995
stages of capacitation are not defined clearly, the lack of HA may relate with the inability of sperm to bind to the zona pellucida and to fertilize an oocyte in vivo. To understand completely the process of capacitation, it was necessary to develop methods of measuring HA in human spermatozoa. Unlike other qualitative measurements, such as sperm morphology, HA involves the motion of spermatozoa and the changes in that motion, making it very difficult and extremely time consuming to measure manually (4, 5). Current technology has combined videomicroscopy and computer-assisted digital image analysis to measure HA (6-9). In this study we compared two computer-assisted sperm analysis (CASA) systems, the newer CellTrak system and the more established Cellsoft system, to assess the reliability of the CellTrak system to assess HA. Hurowitz et a1. Human sperm hyperactivation
427
MATERIALS AND METHODS Sample Preparation
Semen samples were obtained from healthy normal male volunteers after ~48 hours sexual abstinence. Semen was collected by masturbation into sterile wide-mouth plastic containers provided by the laboratory. Samples were allowed to liquefy for 30 minutes at room temperature before processing. Semen was layered onto discontinuous Percoll gradients of80% and 55% (Pharmacia, Uppsala, Sweden), centrifuged, and the final pellet was washed in an N -[2-hydroxyethyl]piperazine-1-[2ethanesulfonic acid]-buffered Ham's F-10 medium with 1% bovine serum albumin (GIBCO, Grand Island, NY). The washed spermatozoa then were suspended at a concentration of 10 to 20 X 1Q6/mL in Ham's F-10 with 10% heat-inactivated pooled human serum obtained from preovulatory women undergoing IVF (capacitation medium) (9). Forty-two sperm preparations were assessed in this study. Sperm Motion Analysis and Hyperactivated Motility
Sperm motion analysis and assessment of HA were performed on spermatozoa incubated in capacitation media after 10 minutes, 1, 5, and 24 hours. Eight microliters of the sperm suspension was placed into 20-J-lm Microcell slides (Fertility Technologies, Natick, MA). All sperm preparations were analyzed on a heated microscope stage (37°C) and recorded by video for ~ 10 minutes. The Cellsoft CASA System (Version 3.51c; Cryo Resources, New York, NY) equipped with a HA module was used for analysis of the videotapes with setup parameters as described previously (7 -9). The videotapes of the sperm preparations were then reanalyzed using the CellTrak CASA System (Version 4.25 Motion Analysis VP 110; Motion Analysis Corporation, Santa Rosa, CA). The following parameters were computed or calculated: curvilinear velocity (VCL, J-lm/s) , straight line velocity (VSL, J-lm/s) , linearity (LIN = VSLNCL X 100), maximum amplitude of lateral head displacement (ALHmru" J-lm), mean amplitude of lateral head displacement (ALHmeam J-lm), and head beat cross frequency (Hz). In addition to the parameters described above, the derived parameters were calculated for each cell: DANCE MEAN = ALHmean X VCLNSL (J-lm). Hyperactivated sperm cells were classified as transitional if the VCL ~ 80 J-lm/s, LIN > 19, and :s; 34 J-lm/s, and DANCEMEAN ~ 17 J-lm, and as star-spin HA if the VCL ~ 80 J-lm/s, LIN :s; 19 J-lm/s, and DANCEMEAN ~ 17 J-lm according to the criteria described previously (6). Total HA is the sum 428
Hurowitz et al. Human sperm hyperactivation
of transitional HA and star-spin HA. Each individual cell track was analyzed ~23 times in 1 second, with a frame rate of 30 frames/so The data from each individual cell track then were recorded and printed. Two hundred to 800 motile cells were analyzed for each sample. Hyperactivated sperm cells were classified and HA data were presented under the criteria as described in this paragraph. Statistical Analyses
Mean, SD, and coefficients of variation (CVs) were computed for several measurements of the same sample by the CellTrak system. The results of all 42 samples for each parameter from each system were compared for agreement using one-factor analysis of variance, linear regression, and the method described by Bland and Altman (10). RESULTS Reproducibility of HA Assessment by CellTrak System
To determine whether the results from the CellTrak system were reproducible, four different samples with extremely low, low, average, and high percentages of hyperactivated spermatozoa (previously determined by the CellSoft system) were analyzed repeatedly by the CellTrak system. The results are summarized in Table 1. The results showed that the sperm motion parameters (percent motility, VCL, VSL, LIN, and ALH) were extremely reproducible, showing low CVs in all parameters. In the HA measurements, the repeated measurements from the high level samples show lower CVs compared with the low level samples. However, even with a mean total HA of 18.8% in sample 4, the CV remained high at 15%. The high CV was due to the small numbers of spermatozoa with HA in each measurement. Counting even higher numbers of sperm might decrease the severity of this variation. The repeated measurements of sample 4 in which 800 motile cells were counted showed further reduction in the CV compared with the measurements where 400 cells were counted for all parameters except one, the percent spermatozoa with transitional HA. The CV for total HA was reduced to <10%. At low HA levels, the system rarely gave a change more than ± 1 cell from the mean number of spermatozoa showing HA; whereas at high levels the computer may yield values up to ±6 cells from the mean. Thus the CellTrak system was much more precise at low levels, but might be less accurate, because ±6 is only 33% of 18 whereas ± 1 is 100% of ± 1 cells. In comparing the highest combination of accuracy and Fertility and Sterility
Table 1 Reproducility of Sperm Motility and HA Parameters as Assessed by the CellTrak CASA System*
Sample It (n
=
13):1:
Sample 2t (n
=
11)
Sample 3t (n
=
8)
Sample 4t (n
=
11)
Sample 4§ (n
=
10)
Percent motility
VCL
VSL
I'm/s
I'm/s
74 ::!: 1.05 (1.43) 80 ::!: 1.64 (2.05) 93 ::!: 1.04 (1.25) 88::!: 1.13 (1.28) 88 ::!: 0.67 (0.76)
70.8 ::!: 2.17 (3.06) 82.6 ::!: 2.45 (2.97) 97.8::!: 3.06 (3.13) 114.8 ::!: 4.88 (4.25) 113.1 ::!: 1.96 (1.73)
48.7 ::!: 2.01 (4.14) 48.1 ::!: 2.20 (4.56) 55 ::!: 2.14 (3.90) 54 ::!: 1.68 (3.10) 53 ::!: 0.94 (1.79)
LIN
ALH I'm
%
%
%
67.8 ::!: 2.43 (3.59) 57.9 ::!: 1.20 (2.08) 60.6 ::!: 2.05 (3.39) 52.1 ::!: 1.51 (2.91) 51.8 ::!: 0.94 (1.82)
3.8 ::!: 0.22 (5.89) 5 ::!: 0.18 (3.65) 5.5 ::!: 0.24 (4.45) 6.8 ::!: 0.28 (4.05) 6.8 ::!: 0.19 (2.77)
0.4 ::!: 0.38 (96.20) 1.6 ::!: 0.88 (55.20) 3.5 ::!: 1.63 (46.73) 7.6::!: 1.63 (21.44) 6.6 ::!: 2.33 (35.48)
0.19 ::!: 0.61 (67.89) 3.9 ::!: 1.29 (32.90) 7.6 ::!: 1.89 (24.83) 11.2 ::!: 1.57 (13.96) 12.3 ::!: 1.49 (12.19)
1.3 ::!: 0.57 (43.63) 5.5 ::!: 1.53 (27.74) 11.1 ::!: 2.10 (18.86) 18.8 ::!: 2.84 (15.09) 18.8 ::!: 1.83 (9.74)
* Values are means::!: SD with the CV, expressed as a percentage in parentheses. t For each parameter, ",,400 motile spermatozoa were counted.
Star-spin HA
Total HA
:I: Number of repeated assessments on the sample. § For each parameter, at ",,800 motile spermatozoa were counted.
preCISIOn samples with HA, between 8% and 10% tend to have the highest combination of accuracy and precision.
DISCUSSION
Computer-assisted sperm analysis systems provide a standardized and unbiased system of measuring and studying human sperm hyperactivated motility and they greatly improve the speed of a slow and tedious manual process (4-9) especially useful in clinical investigations_ Computer and image analysis technology is advancing and CASA technology still is being improved. In our study, even though all of the parameters associated with HA showed small variations when the same sample was analyzed repeatedly, the measurements of transitional HA, star-spin HA, and total HA showed high CVs. This was because low numbers of hyper activated spermatozoa were being counted during each analysis, despite the fact that 400 motile spermatozoa were analyzed for each sample. No sample in this experiment had >20% spermatozoa with HA, which meant that the CellTrak system analyzed <80 hyperactivated spermatozoa per sample. A few more or less spermatozoa counted during each assessment would cause large changes in percent of spermatozoa classified as exhibiting HA, especially in samples with lower levels of HA. Given this limitation, we suggest that as CASA systems become faster and more automated it would be prudent to count ~400 motile
Comparison of HA Assessment of CellTrak With Cellsoft CASA Systems
The values for percent motility, VCL, VSL, LIN, transitional HA, star-spin HA, and total HA for the same 42 samples were measured by each system and compared. Table 2 shows that mean ± SD of the 42 measurements for each parameter by each system as well as linear regression equations describing the relationships of the motion parameters assessed by the two systems. The averages of and the differences between the two assessments of each sample (Cellsoft versus CellTrak) were calculated and plotted for each of the 42 samples (Fig_ 1) to demonstrate the agreement of the results as measured by the two systems_ Good agreement was demonstrated for most of the motion analysis parameters_ The results showed the VSL and VCL measured by the CellTrak system was significantly higher than those measured by the Cellsoft system. The percent of spermatozoa with transitional HA was significantly lower and the percent of spermatozoa with star-spin HA was significantly higher when assessed by the CellTrak when compared with the Cellsoft system_
Table 2
Transitional HA
Comparison of Motion and HA Parameters Measured by the CellTrak and Cellsoft Systems in 42 Semen Preparations Parameter
CellTrak* 81.4 91.4 51.6 58.5 2.6 4.9 7.5
Percent motility VCL (p,m/s) VSL (p,m/s) LIN Transitional HA (%) Star-spin HA (%) Total HA (%)
±
4.41
± 12.1
± ± ± ±
±
7.43 4.91 1.7 2.7 3.7
Cellsoft* 79.4 77.8 46.9 59.7 4.3 2.9 7.2
± 5.3 ± 15.lt ± 1l.6t ±
± ± ±
5.1 2.9t 2.5t 4.1
Equation
r
0.82 0.89 0.85 0.73 0.50 0.36 0.77
CellTrak CellTrak CellTrak CellTrak CellTrak CellTrak CellTrak
= = = = = = =
0.69 0.71 0.54 0.70 0.30 0.38 0.70
x x x x
Cellsoft Cellsoft Cellsoft Cellsoft x Cellsoft x Cellsoft x Cellsoft
+ + + + + + +
26.8 36.1 26.2 16.9 1.31 3.80 2.43
t p < 0.01 when compared with CellTrak.
* Values are means ± SD. Vol. 64, No.2, August 1995
Hurowitz et al.
,
....... " .... . ''''
_ ...
Human sperm hyperactiuation
429
A. % Motility
~
- ==...
'i~
u
u
III
..
U C
..
! u
~ c
B.VCl
4.0 ..,.,..............................~..................-.. ·"'o~.. ·-O~ . ·., 2.0 0.0 -2.0 +-"'O:--:..--"eF-1,-_r-~:F:-:---"---j -4.0 -6.0 -8.0 ·······················. ·o. :250 -10.0 o -12.0 +---_---+----+---~ 85.0 90.0 80.0 70.0 75.0
=
0.0
0 ~
+2S0
Ii:;'" -10.0
u u -II
..
-20.0
~U
-30.0
III
u:!: c: III
~ is
Mean
-2S0
-40.0 60.0
80.0
Average
C.VSL 1~:~
.
-.~--~.-2-s-D'
_~:~ +---=o:.......=(X)Ox-__O__B_~_"G---"O:....:r.Mr-ea=n., -10.0 -15.0
-20.0
8&p
~ ................. r;.
30.0
00
u
- 2S0
......... ..... .................. .
40.0
120.0
D. LIN
"--.-.-..-.-.-.-..-.-..-.-..-.. -. . o
100.0
Average
50.0
60.0
15.0
= 0 ~
10.0
"ii~
o u -II
5.0
u:!: c: III
0.0
III
..
~o
~ is
-10.0 45.0
70.0
00
-5.0 50.0
55.0
Average
60.0
65.0
70.0
Average
E. Transitional Hyperactivity
..
! '""I'ii = .!! 0
'ii
. ..
!:!. u
c
~
it:
i5
F. Star-Spin Hyperactivity 12 10 8 6 4 2 0 -2 -4
o ················;;0
+ 250
······00
u
l!
~
COco
Gj
o
o
o ~(:) .. 8. o
:i'
Mean
~ ;:: 0 ~
- 250
Gj
2
4
6
10
8
2III
u
c:
Average
~
~ is
G. Total Hyperactivity :i' u
l!
~
Gj
~ ;:: 0 ~
Gj
2III
u
c: ~
~ is
6 4 2 0 -2 -4 -6 -8 -10
0
Mean
0 0
-2S0
0 0
5
10
15
Mean
0 0
0 5
10
15
20
Average
430
0
0 0
-2S0
0 0
+2S0
0
0 0
0
+2S0
00 0
0
0 ··········0
Average
"e O " 0
6 4 2 0 -2 -4 -6 -8 -10
Hurowitz et al. Human sperm hyperactiuation
Figure 1 Averages ofthe measurements obtained by the Cell Trak and Cellsoft systems plotted against the differences between them for (Al percent motility, (Bl VCL, (e) VSL, (D) LIN, (El transitional HA, (F) star-spin HA, and (G) total HA. The solid line indicates the mean value of the differences and the dashed lines indicate :!:2 SD from the mean (10).
Fertility and Sterility
spermatozoa to minimize sampling error and the effects of small numbers of spermatozoa with HA on the computer system's analysis of human sperm HA. In comparing the two systems, all of the sperm motion parameters agreed very well except for VCL and the three measurements of HA. In the case of VCL, the CellTrak system consistently measures higher VCL (mean difference 13.7 J-Lrnls) than the Cellsoft system. This may not be an undesirable difference because there is an extremely high correlation between the two measurements of VCL. It is in fact understandable because the CellTrak system uses a more accurate algorithm to track sperm trajectories than the Cellsoft system, especially at higher velocities. The percent cells with transitional HA assessed by CellTrak was systematically lower than that assessed by Cellsoft. The converse was true with the assessment of percent cells with starspin HA. The total percent cells with HA assessed by the two systems was very similar. Because sperm cells move in a whipping and thrashing pattern during HA, a curvilinear pattern during progression as opposed to following a straight line in forward-progressing sperm cells, much of the actual distance the spermatozoa travel is not estimated when considering only how much distance between the two points the sperm traveled. Thus, the closer the CASA system's analyses follow the cell's true path, the higher and more accurate the value of velocity it will give (11). This effect becomes significant in measuring VCL, especially in sperm with a high VCL such those that are classified as star-spin. According to the classification of Robertson et al. (6), the VSL:VCL (LIN) ratio is critical in distinguishing between transitional and star-spin HA patterns. The higher VCL measured by CellTrak system at higher sperm velocities will lower the LIN and classify more spermatozoa as star-spin HA rather than as transitional HA. It should be noted the total HA was comparable as measured by the two systems with no systematic bias. Recognizing the limitations of the CellTrak system and of CASA systems in general (12, 13), we conclude that the technology does effectively provide a quick and unbiased measure of hyperactivity in a semen sample. The CellTrak system showed good reproducibility and it showed a high degree of agreement with the Cellsoft system with most parameters. Beyond limitations common to all CASA systems (12, 13), we found additional problems using CASA systems to measure sperm HA. These are the number of spermatozoa counted and the algorithm used to track sperm trajectories. The former is the reason for the very high variations in transitional HA and star-spin HA measurements. Given those Vol. 64, No.2, August 1995
variations, it is impressive that the measurements of total HA display such a high degree of agreement. Nevertheless, this limitation can be minimized by counting larger numbers of sperm during each sample analysis. The latter limitation introduces a tendency for the computer to underestimate velocity (both VCL and VSL). Thus, the measurements of VCL and VSL estimated by the CellTrak system are probably more accurate than those made by the Cellsoft system. As CASA technology improves and even more accurate approximations to the actual sperm trajectory are developed, such inaccuracies in velocity may decrease, allowing more accurate measurements of HA. The clinical usefulness of measurements of sperm HA has to be examined further in studies to assess the predictive values of these sperm parameters in both assisted and natural human fertilization settings.
Acknowledgments. Evan H. Hurowitz* was a summer high school student supported by the Harbor Collegium Fund. We thank John Greaves, Ph.D., Motion Analysis Corporation (Santa Rosa, California), for responding to our request to develop and allowing us to test the software for assessing HA on the CellTrak CASA system.
REFERENCES 1. Austin CR. Observations on the penetration ofthe sperm into the mammalian egg. Aust J Sci Res B 1951;4:581-96. 2. Austin CR. The 'capacitation' of the mammalian sperm. Nature 1952; 170:326. 3. Yamagamachi R. Mammalian fertilization. In: Knobil E, Neill JD, Ewing LL, Markert CL, Greenwald GS, PfaffDN, editors. The physiology of reproduction. New York: Raven Press, 1988; 131-88. 4. Burkman LJ. Discrimination between nonhyperactivated and classical hyperactivated motility patterns in human spermatozoa using computerized analysis. Fertil Steril 1991; 55:363-71. 5. Mortimer ST, Mortimer D. Kinematics of human spermatozoa incubated under capacitating conditions. J Androl 1990; 11:195-203. 6. Robertson L, WolfDP, Tash JS. Temporal changes in motility parameters related to acrosomal status: identification and characterization of populations of hyperactivated human sperm. BioI Reprod 1988;39:797-805. 7. Wang C, Leung A, Tsoi WL, Leung J, Ng V, Lee KF, et al. Evaluation of human sperm hyperactivated motility and its relationship with the zona-free hamster oocyte sperm penetration assay. J Androl 1991; 12:253-7. 8. Uhler ML, Leung A, Chan SYW, Wang C. Direct effects of
* Acknowledgment from one of the referees: "I find the fact that the lead author was a summer high school student particularly exciting and worthy of special note. We need to encourage our peers to follow your path. Congratulations!" Hurowitz et al. Human sperm hyperactiuation
431
progesterone and antiprogesterone on human sperm hyperactivated motility and acrosome reaction. Fertil Steril 1992; 58:1191-8. 9. Wang C, Lee GS, Leung A, Surrey ES, Chan SYW. Human sperm hyperactivation and acrosome reaction and their relationships to human in vitro fertilization. Fertil Steril 1993;59:1221-7. 10. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307 -10.
432
Hurowitz et al. Human sperm hyperactivation
11. Mortimer D, Serres C, Mortimer ST, Jouannet P. Influence of image sampling frequency on the perceived movement characteristics of progressively motile human spermatozoa. Gamete Res 1988;23:313-27. 12. Boyers SP, Davis RO, Katz DF. Automated semen analysis. Curr Probl Obstet Gynecol Fertil 1989; 12:165-200. 13. Davis RO. The promise and pitfalls of computer-aided sperm analysis. In: Overstreet JW, editor. Male infertility. Infertility and reproductive medicine clinics of North America. Philadelphia: W.B. Saunders Co., 1992:341-52.
Fertility and Sterility