MOLECULAR REPRODUCTION AND DEVELOPMENT 27:326-331(1990)
Prediction of Human Sperm Penetrating Ability Using Computerized Motion Parameters P.M. FETTEROLF AND B.J. ROGERS Center for Fertility and Reproductive Research, Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashville, Tennessee
ABSTRACT The sperm penetration assay (SPA) is used to assess male fertilizing potential but it is tedious and costly. Computer analysis could replace the need for the SPA in some cases, if computerized sperm motility parameters are highly predictive of SPA performance. The objective of this study was to determine whether computerized motility parameters from fresh semen samples could be used to predict SPA performance. Computer automated semen analysis (CASA; CellSoft, Cry0 Resources) was used to quantitate sperm concentration (CONC), percent motility (MOT),curvilinear velocity (VEL), linearity of swimming trajectory (LIN), mean amplitude of lateral head displacement (ALH), and bedcross frequency (B/CF). The SPA was performed using either Biggers, Whitten, and Whittingham’s medium (BWW) or TEST-yolk buffer (TYB). Patients were divided into three groups depending on SPA performance: group 1, B W - t r e a t e d , 0% versus > 0% penetration; group 2, TYB-treated, 0% versus > 0% penetration; and group 3, TYB-treated, < 20% versus 2 20% penetration. SPA performance was highly correlated with CONC, MOT, VEL, and B/CF. CONC, MOT, VEL, and B/CF were significantly higher for patients who penetrated in the SPA than far those who failed to penetrate. Discriminant function analysis (DFA) successfully classified 76% of all patients treated with TYB (group 2) who penetrated and 86% of nonpenetrators based on their computerized motility parameters. For group 2 DFA predicted that 93 men would penetrate in the SPA with TYB. Of these, 90 (97%)successfully penetrated at least one egg. It is concluded that DFA of computerized motility parameters can accurately predict whether a patient will penetrate (> 0%) in the TYB-treated SPA.
eggs (Yanagamachi et al., 1976; Rogers et al., 1979; Rogers, 1983). This widely used assay of human sperm function provides a tool for assessing the fertilizing potential of the male in infertile couples (Overstreet et al., 1980; Karp et al., 1981; Aitken e t al., 1982; Rogers et al., 1985a,b; Smith et al., 1987). The SPA is tedious and costly. Consequently, it would be advantageous to develop a more rapid, less costly method for predicting the ability of human sperm to penetrate eggs. The objective of this study was to determine if computerized motility parameters from fresh semen samples can be used to predict hamster egg penetration in the sperm penetration assay. The relationship between sperm motility parameters and fertilizing potential has been investigated by several researchers. Using different approaches, three studies (Aitken et al., 1982; Mathur et al., 1986; Milligan et al., 1980) have suggested that sperm swimming speed is a n important variable in discriminating between fertile and infertile individuals. In contrast, Hope and co-workers (Hope et al., 1984) found no such relationship. Aitken e t al. (19821, Holt e t al. (19851, and Ginsberg and his colleagues (1987) used penetration of zona-free hamster eggs to measure fertilizing potential. Aitken and his colleagues (1982) used timeexposure photomicrography to quantitate motility parameters. They reported that straight swimming mode of progression, small lateral head displacement value, and rapid swimming speed were positively correlated with penetration rate into zona-free hamster eggs. Similarly, Holt and co-workers (1985) used a semiautomatic computerized technique and found that sperm swimming speed was correlated with penetration rate in the SPA. Using a multivariate statistical approach that simultaneously analyzed all motility parameters,
Key Words: Computer assisted semen analysis (CASA), Sperm penetration assay (SPA), Human sperm fertilizing potential
INTRODUCTION The sperm penetration assay (SPA) quantitates the ability of human sperm to penetrate zona-free hamster
0 1990 WILEY-LISS, INC.
Received January 16, 1990; accepted May 30, 1990. Address reprint requests to Dr. B. Jane Rogers, Department of OBi GYN, C-FARR, D-3224, Medical Center North, Vanderbilt University Medical, Nashville, TN 37232. Dr. Fetterolf is now at Division of Reproductive Sciences, CCRW3815, The Toronto Hospital, Toronto General Division, 200 Elizabeth Street, Toronto, Ontario, Canada M5G 2C4.
SPA AND MOTILITY PARAMETERS
327
THE SPERM PENETRATION ASSAY Sperm Preparation
Zona-Free Egg Preparation
Semen Collection (Day 3) Hamster Injected with PMS (Day 1) and hCG (Day 3)
Two Washes with BWW
$-
Y CaDacitatwn in TYB
Add Equal VoCma of TYB to Suspension
Incubate i r Hrs. at 40
Capacitatwn in B w w
Two Washes wnh BWW. Adjust Sperm Concentration
$-
Add BWW to Adjust Sperm Concentration
1
c
T h s and Cumulus Removed (Day 4)
%
b
Incubate 1 Hr. at 3 7 O C
+
0 0 $-
Incubate 18 Hrs. at 37O C
Sperm and Eggs Incubated for 2 Hours
Cumulus Dispersed with Hyalwonidase
ZonaRemoved with Trypsin
Y
E m Examined for Sperm Penetration
Fig. 1. Schematic drawing of the sperm penetration assay protocol showing the BWW and TYB treatments.
they were able to predict the rate of hamster egg penetration for only 75% of the study subjects. Ginsberg et al. (1987) measured sperm motion with a videomicrographic system (CellSoft, Cryo Resources) and attempted to relate these measurements to penetration in the SPA. Multivariate statistical analysis allowed them to predict hamster egg penetration with a n average of 71% accuracy based on motility parameters. Computer automated semen analysis (CASA; CellSoft) provides rapid, objective measurement of sperm concentration (CONC), percent motility (MOT), curvilinear velocity (VEL), linearity of swimming trajectory (LIN), mean amplitude of lateral head displacement (ALH), and beatkross frequency (B/CF). CASA-generated motility parameters could be used in place of the SPA for many patients if these motility parameters, obtained from fresh semen specimens, are highly predictive of egg-penetrating ability in the SPA.
MATERIALS AND METHODS Sperm Penetration Assay Fresh semen samples were collected by masturbation from 147 infertility patients. After liquefaction, a n aliquot was removed for computerized motility analysis and the remainder used in the SPA. The SPA sample was treated using two protocols: the standard SPA using BWW (Method 1) and the modified SPA using TEST-yolk buffer (Method 2). The methods have been previously described (Rogers, 1985) and are summarized briefly. The first step in both methods was three
washes with BWW, followed by resuspension of the sperm pellet in 1 ml BWW (Fig. 1).
SPA Method 1 (BWW, Fig. 1) A portion of the sperm suspension was diluted to 10 x 106/ml with BWW and 0.5 ml aliquots were stored overnight in tightly capped polystyrene tubes a t 37°C in a n air incubator. This is called the BWW sample. SPA Method 2 (TYB, Fig. 1) The remainder of the sperm suspension was diluted with a n equal volume of TEST-yolk buffer (TYB) and stored overnight at 4°C. This is called the TYB sample. The following morning, the TYB sample was warmed at 37°C for 1h r in a n air incubator, washed twice with BWW to remove the TEST-yolk buffer, and the sperm diluted to 10 x 106/ml. Eggs were obtained from superovulated hamsters, the cumulus oophorus was removed with hyaluronidase (Sigma H-3506, Sigma Chemical Co., St. Louis, MO), and the zona pellucida was removed using trypsin (Sigma T-0134). Drops (100 p1) of BWW- and TYBtreated sperm (10 x 106/ml) were placed under mineral oil in Petri dishes and a t least 30 zona-free hamster eggs were pipetted into the sperm droplet. After a 2-3 h r incubation, the number of eggs penetrated and the number of swollen sperm heads in each hamster egg were counted. Two parameters were quantitated in the SPA. Percent penetration (%P) is the percentage of hamster
328
P.M. FETTEROLF AND B.J. ROGERS
eggs containing at least one swollen sperm head. Penetration index (PI) is the average number of swollen sperm heads per hamster egg.
CASA Analysis The CellSoft software is supported by a n IBM Turbo AT computer (International Business Machines, Boca Raton, FL), a Panasonic KX-P1091 Dot Matrix Printer (Matsushita Communication Industrial Co., Ltd., Tokyo, Japan), a n Olympus BH-2 phase-contrast microscope (Olympus Optical Co., Ltd., Tokyo, Japan) with 10 x objective, a Panasonic high-resolution video camera (model WV-1410), two Panasonic monitors (model TR-930), and a Makler counting chamber (Israel Electrotypical Industry Ltd., Rehovot, Israel). Semen samples used for computer analysis were diluted to approximately 25 x lo6 sperm/ml using Biggers, Whitten, and Whittingham’s medium (BWW). Dilution of high concentration samples is necessary to optimize the accurate determination of sperm motility parameters by CellSoft (Critser et al., 1988; Vantman et al., 1988). Seven microliters of the sample were placed in a Makler counting chamber and the coverslip was pressed down firmly to insure a uniform 10 pm layer of sample. At least 200 sperm were analyzed for each sample. Samples with fewer than 5 x lo6 sperm/ml were not analyzed. Also, samples with numerous, large (> 20 pm) debris were not analyzed because these pieces of foreign matter were surrounded by bright “halos”, which the computer incorrectly interpreted as sperm motion tracks. The computer settings used for the study were as follows: minimum samples motile, 4; threshold velocity, 8 pm/sec; minimum samples velocity, 20; size (low, high) 5, 25; maximum velocity, 300 pmhec; and pixel scale, 0.688. Statistical Analysis Relationships between semen parameters and %P and PI for BWW and TYB samples were determined using Spearman rank correlation. For further analysis, patients were divided into three groups based on SPA performance: (group 1)BWW-treated, 0% vs. > 0% penetration in SPA; (group 2) TYB-treated, 0% vs. > 0% penetration in SPA; and (group 3) TYB- treated, < 20% vs. 2 20% penetration in SPA. The breakdown of patients in groups for statistical analysis was based on previously published information and on data generated in our laboratory. I t has been suggested that 0% penetration in the BWW-treated SPA indicates potentially infertile men and that penetration of at least one egg indicates potentially fertile men (Rogers, 1985; Rogers, 1986). Thus, 0% penetration and > 0% penetration were considered as the two groups in the BWWtreated SPA. However, re-evaluation of a larger data pool showed a false negative rate of approximately 30% in the BWW-treated SPA. This false negative rate appears to be eliminated by the TYB-treated SPA (Falk et al., 1990). With TYB, the normal percent penetra-
TABLE 1. Spearman Correlations of Objective Semen Parameters and SPA Performance With BWW PI
% Pen
Concentration ( X 106/ml) Motility (%) Velocity (kmlsec) Linearity Mean ALH BIC frequency
0.39*** 0.40*** 0.40*** 0.26**
0.16 0.35***
0.37*** 0.39*** 0.41*** 0.24* 0.18 0.33***
With TYB PI
% Pen
0.48*** 0.54*** 0.45*** 0.19*
0.48*** 0.55*** 0.45***
0.28**
0.20* 0.28**
0.33***
0.33***
*P < 0.05. **P < 0.01. ***P < 0.001.
tion for fertile men has been suggested as 20% or greater (Falk et al., 1990), so 20% has been used as a cutoff for statistical analysis. A few men achieving pregnancies fell into the 10-20% penetration range in the TYB-treated SPA and were considered as subfertile (Falk et al., 1990). The 0% vs. > 0% group for penetration in the TYB-treated SPA was chosen because the strict 0% threshold appears to classify men a s either fertile or infertile. Sample sizes vary among groups because some individuals were tested with only BWW or TYB. Objective motility parameters were compared for penetration categories within these groups using t tests and discriminant function analysis (DFA). DFA is a multivariate statistical procedure which simultaneously analyzes all variables to generate a discriminant equation and then casts each sample into the prescribed categories (e.g., 0% vs. > 0% penetration) based on the equation. The discriminant equation maximizes the differences between samples in order to generate distinct groups. If the sample groups generated by the equation exactly match the prescribed penetration categories, then the equation can be used on any sample to predict its penetration category. Statistical significance was for P < 0.05. Statistical analysis was performed using the Statistical Analysis System for Personal Computers (SAS), Version 6 (SAS Institute Inc., Cary, NC).
RESULTS Correlation Analysis Strong, significant correlations were found between CASA-generated semen parameters and SPA performance for samples treated with BWW and TYB (Table 1). Velocity, beatkross frequency, percent motility, and sperm concentration were highly positively correlated with egg penetration. There were two notable differences between BWW and TYB treatment. The first was that correlations between percent penetration and objective semen parameters were higher with TYB (n = 140) than with BWW (n = 131) for CONC (0.48 vs. 0.391, MOT (0.54 vs. 0.401, VEL (0.45 vs. 0.40), and ALH (0.28 vs. 0.16). In contrast, the relationship be-
SPA AND MOTILITY PARAMETERS BWW Lateral head ttt
om )o%
0% ,05
ttt
I n 5 -
TYB 3 1 sacepee. ~ ‘veaI
I
***
329 TYB -. tt
,206 1.20%
Percent Penetration
85
Linearlty
e
3%
)om
0% j0P
Ferce-! Peietra:lo*
(20% ).20%
0%
22
I
t
SO%
Beat/Cross Frecjuercy
,
204
0%
,om
***
*tt
. t
120s 1.20%
Percent Penetratlon
Fig. 2. Relationship of motility parameters to penetration status. Bar graph showing the computerized semen characteristics (mean ? SE) of patients in different groups based on percent penetration in the sperm penetration assay. Significant differences between penetration categories are shown (*** P < 0.001,** P < 0.01,* P < 0.05).
tween percent penetration and LIN or B/CF did not exhibit this trend. The correlations between penetration index and CASA-generated semen parameters showed the same pattern as for percent penetration. The second difference between BWW and TYB treatment was that ALH was significantly correlated with SPA performance for samples treated with TYB but not those treated with BWW.
SPA Performance Groups-t Tests Velocity, beat/cross frequency, motility, concentration, and linearity were all significantly higher for penetrators (> 0% penetration, n = 76) than for nonpenetrators (0% penetration, n = 55) in the BWW-treated SPA (group 1, Fig. 2). Similarly, VEL, ALH, B E F , MOT, and CONC were significantly higher for penetrators (n = 119) than for nonpenetrators (n = 21) in the TYB-treated SPA (group 2). In group 3, VEL, ALH, BC/F, MOT, and CONC were significantly higher for high penetrators ( 2 20% penetration, n = 74) than for low penetrators (< 20% penetration, n = 66). Differences between penetration categories within groups were substantial for several motility parameters (e.g., group 2, 69.6 vs. 53.3 pm/sec for velocity) with the most pronounced differences in group 2 (Fig. 2). Nevertheless, the variation around the mean was high and the range of values overlapped considerably. The most notable difference between BWW- and TYBtreated SPA performance was that mean ALH was significantly different between penetration categories with TYB but not with BWW. Patients who penetrated
better in the SPA had higher ALH in their fresh samples.
SPA Performance Groups-Discriminant Function Analysis DFA correctly classified the SPA performance of 67% (51 out of 76) of the penetrators and 71% (39 out of 55) of the nonpenetrators from the CASA motility parameters of patients treated with BWW (group 1). For patients treated with TYB (group 2), DFA correctly classified the SPA performance of 76% (90 out of 119) of the penetrators and 86% (18 out of 21) of the nonpenetrators from their CASA motility parameters. The level of penetration for group 3 was similar to group 1 with DFA correctly classifying the SPA performance of 68% (53 out of 78) of the high penetrators and 68% (47 out of 69) of the low penetrators. For all groups, DFA classified significantly more SPA results correctly than if the result had been chosen by chance. However, percent correct classification was relatively high only for group 2. Motion parameters added significant information to the discriminant analysis. For example, in TYB with 0% penetration a s cutoff (group 2) inclusion of volume, concentration, and percent motility alone in the discriminant analysis produced only 67% correct classification compared to 81% when motion parameters were included. Also in this group, only 67% of nonpenetrators were correctly predicted with conventional semen parameters whereas 86% were correctly predicted when motion parameters were included in the DFA.
330
P.M. FETTEROLF AND B.J. ROGERS
Using CASA-generated motility parameters for group 2 patients, DFA predicted that 93 men would penetrate in the SPA with TYB. Of these patients, 90 (97%) successfully penetrated a t least one egg. The three men who failed to penetrate initially were able to penetrate in a subsequent SPA test or with BWW. Thus, the discriminant function equation can be used to accurately predict TYB-treated SPA penetrators (> 0%) from computerized motility parameters.
was comparably successful at classifying the SPA performance of patients based on motility parameters. In the best case (TYB-treated SPA, 0% vs. > 0% penetration), DFA correctly classified 76% of the penetrators and 86% of the nonpenetrators. Such levels of classification would seem to preclude the use of the DFA and motility parameters to replace the SPA. However, the discriminant function equation predicted that 93 patients would penetrate with TYB treatment and 90 (97%)succeeded. Of the three men who failed, two penetrated in subsequent SPAs and the other penetrated in the BWW-treated SPA. Thus, this discriminant equation can be used to accurately predict future successful penetrators from computerized motility characteristics. This discriminant function could have saved 63% (93 out of 147) of our patients the expense of the SPA, if a penetration threshold of 0% for the TYBtreated SPA is considered diagnostic in the infertility workup. It is concluded that objective motility parameters can predict hamster egg penetration (> 0%) in the TYBtreated SPA using multivariate statistical analysis (DFA). This approach appears to be a potentially useful, rapid alternative to the SPA for predicting sperm penetrating ability in a large subset of infertility patients.
DISCUSSION Some investigators have reported significant relationships between objectively determined motility parameters and SPA performance (Aitken et al., 1985; Holt et al., 1985; Ginsberg et al., 1987) but others have found no relationship (Kossoy et al., 1987). Results reported in this study support the hypothesis that the vigor of sperm motion is positively related to quantitative measures of penetration of zona-free hamster eggs. Measures of sperm motion such as velocity, beat-cross frequency and percent motility were all highly significantly (P < 0.001) related to penetration. In addition, these same parameters were significantly higher (P < 0.001) for relatively high penetrators than for low or unsuccessful penetrators when compared within the SPA performance groups. Unfortunately, the correlations and penetration-related differences in motility ACKNOWLEDGMENTS parameters reported herein are not sufficiently strong to permit prediction of SPA outcome with confidence. We thank Tom Thompson and Kay Gibson for perPrevious investigations that have found no relation- forming the SPAs and Kathy Carter and Charlene ship between computerized motility parameters and Thompson for help in preparation of the manuscript. SPA performance may have been confounded by lack of dilution of the semen sample (Critser et al., 1988; REFERENCES Vantman et al., 1988) or inappropriate settings for the Aitken RJ,Best FSM, Richardson DW, Djahanbakch D, Mortimer D, Templeton AA, Lees MM (1982): An analysis of sperm function in computer software (Knuth et al., 1987; Mack e t al., cases of unexplained infertility: Conventional criteria, movement 1988). It has been demonstrated that such factors can characteristics, and fertilizing capacity. Fertil Steril 38:212-2 19. greatly affect computerized motility analysis (Critser Critser JK, Colvin KE, Critser ES (1988): Effect of sperm concentraet al., 1988; Vantman et al., 1988; Knuth et al., 1987; tion on computer assisted semen analysis results. Andrology 9:14Mack et al., 1988). Using the same computer system a s P. in this study, Kossoy et al. (1987) found no strong re- Falk RM, Silverberg KM, Fetterolf PM, Kirchner FK, Rogers BJ (1990):Establishment of TEST-yolk buffer enhanced SPA limits for lationships between motility parameters and SPA perfertile males. Fertil Steril 54:121-126. formance. In that study, semen samples were not di- Ginsberg KA, Moghissi KS, Abel EL, Ager JW (1987): Predicting luted and computer settings were not optimal (Knuth hamster egg penetration by computer-assisted sperm movement analysis. Fertil Steril 48:40. et al., 1987; Mack et al., 1988). In comparison, samples in this study were diluted to a concentration within the Holt WV, Moore HDM, Hillier SG (1985): Computer assisted measurement of sperm swimming speed in human semen: Correlation optimal range for CellSoft (Vantman et al., 1988) and of results with in vitro fertilization assays. Fertil Steril44:112-119. the computer settings were very similar to those which Hope E, Blackburn K, Zenick H, Rafales L, Meyer C (1984): Computhave been determined to be optimal (Mack et al., 1988). erized evaluation of human sperm. Biol Reprod 30:176. In a n unpublished study similar to this one, Gins- Karp LE, Williamson RA, Moore De, Shy KK, Plymate R, Smith WD (1981): Sperm penetration assay: Useful test in evaluation of male berg and colleagues (1987) compared computer-generfertility. Obstet Gynecol 57:620-623. ated motility parameters with percent penetration in Knuth UA, Yeung C, Nieschlag E (1987): Computerized semen analthe BWW-treated SPA. They divided patients into 0ysis: Objective measurement of semen characteristics is biased by 10% and > 10% penetration groups and reported that subjective parameter setting. Fertil Steril 48:118-124. discriminant function analysis correctly classified 81% Kossoy LR, Thompson T, Buck A, Hinson M, Brodie B, Vaughn W, Wentz AC, Rogers BJ (1987): Comparison of the sperm penetration of the samples in the low penetration group and 60% of assay (SPA) with objective motility parameters. Andrology 8:38-P. samples in the high penetration group. In our study Mack SO, Wolf DP, Tash JS (1988): Quantitation of specific paramewith a larger sample, different SPA methodologies, and ters of motility in large numbers of human sperm by digital image processing. Biol Reprod 38:270-281. different group cutoffs, discriminant function analysis
SPA AND MOTILITY PARAMETERS Mathur S, Carlton M, Ziegler J , Rust PF, Williamson HO (1986): A computerized sperm motion analysis. Fertil Steril 46:484-488. Milligan MP, Harris S, Dennis KJ (1980): Comparison of sperm velocity in fertile and infertile groups as measured by time lapse photography. Fertil Steril 34:509-511. Overstreet JW, Yanagamachi R, Katz DF, Hayashi K, Hanson FW (1980): Penetration of human spermatozoa into the human zona pellucida and the zona-free hamster egg: A study of fertile donors and infertile patients. Fertil Steril 33534-542. Rogers BJ (1983): Hamster egg: evaluation of human sperm using in vitro fertilization. In PG Crosignai (ed): “In Vitro Fertilization and Embryo Transfer, Proceedings of the Serono Symposium.” New York: Academic Press, p 101. Rogers BJ (1985):The sperm penetration assay: Its usefulness reevaluated. Fertil Steril 43:821-840. Rogers BJ (1986): The usefulness of the sperm penetration assay in predicting in vitro fertilization (IVF) success. J Vitro Fertil Embryo Transfer 3:209 -2 11. Rogers BJ, Holmgren W, Krivacka T, Maxson WS, Wentz AC (1985a):
331
Enhancement of fertilizing capacity by TEST-yolk buffer. 41st Annual Meeting of the American Fertility Society, Chicago, Illinois. The Program Supplement, p. 46, number 83. Rogers BJ, Mygatt GG, Soderdahl DW, Hale RW (1985b): Monitoring of suspected infertile men with varicocele by the sperm penetration assay. Fertil Steril 44:800-805. Rogers BJ, Van Campen H, Ueno M, Lambert H, Bronson R, Hale R (1979): Analysis of human spermatozoa1 fertilizing ability using zona-free ova. Fertil Steril 32:664-670. Smith RG, Johnson A, Lamb D, Lipshultz LI (1987): Functional tests of spermatozoa. Urol Clin N Am 14:451-458. Vantman D, Koukoulis G, Dennison L, Zinaman M, Sherins R J (1988): Computer-assisted semen analysis: Evaluation of method and assessment of the influence of sperm concentration on linear velocity determination. Fertil Steril 49:510-515. Yanagamachi R, Yanagamachi H, Rogers BJ (1976): The use of zonafree animal ova as a test system for the assessment of the fertilizing capacity of human spermatozoa. Biol Reprod 15:471-476.
Prediction of Human Sperm Penetrating Ability Using Computerized Motion Parameters P.M. FETTEROLF AND B.J. ROGERS Center for Fertility and Reproductive Research, Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashville, Tennessee
ABSTRACT The sperm penetration assay (SPA) is used to assess male fertilizing potential but it is tedious and costly. Computer analysis could replace the need for the SPA in some cases, if computerized sperm motility parameters are highly predictive of SPA performance. The objective of this study was to determine whether computerized motility parameters from fresh semen samples could be used to predict SPA performance. Computer automated semen analysis (CASA; CellSoft, Cry0 Resources) was used to quantitate sperm concentration (CONC), percent motility (MOT),curvilinear velocity (VEL), linearity of swimming trajectory (LIN), mean amplitude of lateral head displacement (ALH), and bedcross frequency (B/CF). The SPA was performed using either Biggers, Whitten, and Whittingham’s medium (BWW) or TEST-yolk buffer (TYB). Patients were divided into three groups depending on SPA performance: group 1, B W - t r e a t e d , 0% versus > 0% penetration; group 2, TYB-treated, 0% versus > 0% penetration; and group 3, TYB-treated, < 20% versus 2 20% penetration. SPA performance was highly correlated with CONC, MOT, VEL, and B/CF. CONC, MOT, VEL, and B/CF were significantly higher for patients who penetrated in the SPA than far those who failed to penetrate. Discriminant function analysis (DFA) successfully classified 76% of all patients treated with TYB (group 2) who penetrated and 86% of nonpenetrators based on their computerized motility parameters. For group 2 DFA predicted that 93 men would penetrate in the SPA with TYB. Of these, 90 (97%)successfully penetrated at least one egg. It is concluded that DFA of computerized motility parameters can accurately predict whether a patient will penetrate (> 0%) in the TYB-treated SPA.
eggs (Yanagamachi et al., 1976; Rogers et al., 1979; Rogers, 1983). This widely used assay of human sperm function provides a tool for assessing the fertilizing potential of the male in infertile couples (Overstreet et al., 1980; Karp et al., 1981; Aitken e t al., 1982; Rogers et al., 1985a,b; Smith et al., 1987). The SPA is tedious and costly. Consequently, it would be advantageous to develop a more rapid, less costly method for predicting the ability of human sperm to penetrate eggs. The objective of this study was to determine if computerized motility parameters from fresh semen samples can be used to predict hamster egg penetration in the sperm penetration assay. The relationship between sperm motility parameters and fertilizing potential has been investigated by several researchers. Using different approaches, three studies (Aitken et al., 1982; Mathur et al., 1986; Milligan et al., 1980) have suggested that sperm swimming speed is a n important variable in discriminating between fertile and infertile individuals. In contrast, Hope and co-workers (Hope et al., 1984) found no such relationship. Aitken e t al. (19821, Holt e t al. (19851, and Ginsberg and his colleagues (1987) used penetration of zona-free hamster eggs to measure fertilizing potential. Aitken and his colleagues (1982) used timeexposure photomicrography to quantitate motility parameters. They reported that straight swimming mode of progression, small lateral head displacement value, and rapid swimming speed were positively correlated with penetration rate into zona-free hamster eggs. Similarly, Holt and co-workers (1985) used a semiautomatic computerized technique and found that sperm swimming speed was correlated with penetration rate in the SPA. Using a multivariate statistical approach that simultaneously analyzed all motility parameters,
Key Words: Computer assisted semen analysis (CASA), Sperm penetration assay (SPA), Human sperm fertilizing potential
INTRODUCTION The sperm penetration assay (SPA) quantitates the ability of human sperm to penetrate zona-free hamster
0 1990 WILEY-LISS, INC.
Received January 16, 1990; accepted May 30, 1990. Address reprint requests to Dr. B. Jane Rogers, Department of OBi GYN, C-FARR, D-3224, Medical Center North, Vanderbilt University Medical, Nashville, TN 37232. Dr. Fetterolf is now at Division of Reproductive Sciences, CCRW3815, The Toronto Hospital, Toronto General Division, 200 Elizabeth Street, Toronto, Ontario, Canada M5G 2C4.
SPA AND MOTILITY PARAMETERS
327
THE SPERM PENETRATION ASSAY Sperm Preparation
Zona-Free Egg Preparation
Semen Collection (Day 3) Hamster Injected with PMS (Day 1) and hCG (Day 3)
Two Washes with BWW
$-
Y CaDacitatwn in TYB
Add Equal VoCma of TYB to Suspension
Incubate i r Hrs. at 40
Capacitatwn in B w w
Two Washes wnh BWW. Adjust Sperm Concentration
$-
Add BWW to Adjust Sperm Concentration
1
c
T h s and Cumulus Removed (Day 4)
%
b
Incubate 1 Hr. at 3 7 O C
+
0 0 $-
Incubate 18 Hrs. at 37O C
Sperm and Eggs Incubated for 2 Hours
Cumulus Dispersed with Hyalwonidase
ZonaRemoved with Trypsin
Y
E m Examined for Sperm Penetration
Fig. 1. Schematic drawing of the sperm penetration assay protocol showing the BWW and TYB treatments.
they were able to predict the rate of hamster egg penetration for only 75% of the study subjects. Ginsberg et al. (1987) measured sperm motion with a videomicrographic system (CellSoft, Cryo Resources) and attempted to relate these measurements to penetration in the SPA. Multivariate statistical analysis allowed them to predict hamster egg penetration with a n average of 71% accuracy based on motility parameters. Computer automated semen analysis (CASA; CellSoft) provides rapid, objective measurement of sperm concentration (CONC), percent motility (MOT), curvilinear velocity (VEL), linearity of swimming trajectory (LIN), mean amplitude of lateral head displacement (ALH), and beatkross frequency (B/CF). CASA-generated motility parameters could be used in place of the SPA for many patients if these motility parameters, obtained from fresh semen specimens, are highly predictive of egg-penetrating ability in the SPA.
MATERIALS AND METHODS Sperm Penetration Assay Fresh semen samples were collected by masturbation from 147 infertility patients. After liquefaction, a n aliquot was removed for computerized motility analysis and the remainder used in the SPA. The SPA sample was treated using two protocols: the standard SPA using BWW (Method 1) and the modified SPA using TEST-yolk buffer (Method 2). The methods have been previously described (Rogers, 1985) and are summarized briefly. The first step in both methods was three
washes with BWW, followed by resuspension of the sperm pellet in 1 ml BWW (Fig. 1).
SPA Method 1 (BWW, Fig. 1) A portion of the sperm suspension was diluted to 10 x 106/ml with BWW and 0.5 ml aliquots were stored overnight in tightly capped polystyrene tubes a t 37°C in a n air incubator. This is called the BWW sample. SPA Method 2 (TYB, Fig. 1) The remainder of the sperm suspension was diluted with a n equal volume of TEST-yolk buffer (TYB) and stored overnight at 4°C. This is called the TYB sample. The following morning, the TYB sample was warmed at 37°C for 1h r in a n air incubator, washed twice with BWW to remove the TEST-yolk buffer, and the sperm diluted to 10 x 106/ml. Eggs were obtained from superovulated hamsters, the cumulus oophorus was removed with hyaluronidase (Sigma H-3506, Sigma Chemical Co., St. Louis, MO), and the zona pellucida was removed using trypsin (Sigma T-0134). Drops (100 p1) of BWW- and TYBtreated sperm (10 x 106/ml) were placed under mineral oil in Petri dishes and a t least 30 zona-free hamster eggs were pipetted into the sperm droplet. After a 2-3 h r incubation, the number of eggs penetrated and the number of swollen sperm heads in each hamster egg were counted. Two parameters were quantitated in the SPA. Percent penetration (%P) is the percentage of hamster
328
P.M. FETTEROLF AND B.J. ROGERS
eggs containing at least one swollen sperm head. Penetration index (PI) is the average number of swollen sperm heads per hamster egg.
CASA Analysis The CellSoft software is supported by a n IBM Turbo AT computer (International Business Machines, Boca Raton, FL), a Panasonic KX-P1091 Dot Matrix Printer (Matsushita Communication Industrial Co., Ltd., Tokyo, Japan), a n Olympus BH-2 phase-contrast microscope (Olympus Optical Co., Ltd., Tokyo, Japan) with 10 x objective, a Panasonic high-resolution video camera (model WV-1410), two Panasonic monitors (model TR-930), and a Makler counting chamber (Israel Electrotypical Industry Ltd., Rehovot, Israel). Semen samples used for computer analysis were diluted to approximately 25 x lo6 sperm/ml using Biggers, Whitten, and Whittingham’s medium (BWW). Dilution of high concentration samples is necessary to optimize the accurate determination of sperm motility parameters by CellSoft (Critser et al., 1988; Vantman et al., 1988). Seven microliters of the sample were placed in a Makler counting chamber and the coverslip was pressed down firmly to insure a uniform 10 pm layer of sample. At least 200 sperm were analyzed for each sample. Samples with fewer than 5 x lo6 sperm/ml were not analyzed. Also, samples with numerous, large (> 20 pm) debris were not analyzed because these pieces of foreign matter were surrounded by bright “halos”, which the computer incorrectly interpreted as sperm motion tracks. The computer settings used for the study were as follows: minimum samples motile, 4; threshold velocity, 8 pm/sec; minimum samples velocity, 20; size (low, high) 5, 25; maximum velocity, 300 pmhec; and pixel scale, 0.688. Statistical Analysis Relationships between semen parameters and %P and PI for BWW and TYB samples were determined using Spearman rank correlation. For further analysis, patients were divided into three groups based on SPA performance: (group 1)BWW-treated, 0% vs. > 0% penetration in SPA; (group 2) TYB-treated, 0% vs. > 0% penetration in SPA; and (group 3) TYB- treated, < 20% vs. 2 20% penetration in SPA. The breakdown of patients in groups for statistical analysis was based on previously published information and on data generated in our laboratory. I t has been suggested that 0% penetration in the BWW-treated SPA indicates potentially infertile men and that penetration of at least one egg indicates potentially fertile men (Rogers, 1985; Rogers, 1986). Thus, 0% penetration and > 0% penetration were considered as the two groups in the BWWtreated SPA. However, re-evaluation of a larger data pool showed a false negative rate of approximately 30% in the BWW-treated SPA. This false negative rate appears to be eliminated by the TYB-treated SPA (Falk et al., 1990). With TYB, the normal percent penetra-
TABLE 1. Spearman Correlations of Objective Semen Parameters and SPA Performance With BWW PI
% Pen
Concentration ( X 106/ml) Motility (%) Velocity (kmlsec) Linearity Mean ALH BIC frequency
0.39*** 0.40*** 0.40*** 0.26**
0.16 0.35***
0.37*** 0.39*** 0.41*** 0.24* 0.18 0.33***
With TYB PI
% Pen
0.48*** 0.54*** 0.45*** 0.19*
0.48*** 0.55*** 0.45***
0.28**
0.20* 0.28**
0.33***
0.33***
*P < 0.05. **P < 0.01. ***P < 0.001.
tion for fertile men has been suggested as 20% or greater (Falk et al., 1990), so 20% has been used as a cutoff for statistical analysis. A few men achieving pregnancies fell into the 10-20% penetration range in the TYB-treated SPA and were considered as subfertile (Falk et al., 1990). The 0% vs. > 0% group for penetration in the TYB-treated SPA was chosen because the strict 0% threshold appears to classify men a s either fertile or infertile. Sample sizes vary among groups because some individuals were tested with only BWW or TYB. Objective motility parameters were compared for penetration categories within these groups using t tests and discriminant function analysis (DFA). DFA is a multivariate statistical procedure which simultaneously analyzes all variables to generate a discriminant equation and then casts each sample into the prescribed categories (e.g., 0% vs. > 0% penetration) based on the equation. The discriminant equation maximizes the differences between samples in order to generate distinct groups. If the sample groups generated by the equation exactly match the prescribed penetration categories, then the equation can be used on any sample to predict its penetration category. Statistical significance was for P < 0.05. Statistical analysis was performed using the Statistical Analysis System for Personal Computers (SAS), Version 6 (SAS Institute Inc., Cary, NC).
RESULTS Correlation Analysis Strong, significant correlations were found between CASA-generated semen parameters and SPA performance for samples treated with BWW and TYB (Table 1). Velocity, beatkross frequency, percent motility, and sperm concentration were highly positively correlated with egg penetration. There were two notable differences between BWW and TYB treatment. The first was that correlations between percent penetration and objective semen parameters were higher with TYB (n = 140) than with BWW (n = 131) for CONC (0.48 vs. 0.391, MOT (0.54 vs. 0.401, VEL (0.45 vs. 0.40), and ALH (0.28 vs. 0.16). In contrast, the relationship be-
SPA AND MOTILITY PARAMETERS BWW Lateral head ttt
om )o%
0% ,05
ttt
I n 5 -
TYB 3 1 sacepee. ~ ‘veaI
I
***
329 TYB -. tt
,206 1.20%
Percent Penetration
85
Linearlty
e
3%
)om
0% j0P
Ferce-! Peietra:lo*
(20% ).20%
0%
22
I
t
SO%
Beat/Cross Frecjuercy
,
204
0%
,om
***
*tt
. t
120s 1.20%
Percent Penetratlon
Fig. 2. Relationship of motility parameters to penetration status. Bar graph showing the computerized semen characteristics (mean ? SE) of patients in different groups based on percent penetration in the sperm penetration assay. Significant differences between penetration categories are shown (*** P < 0.001,** P < 0.01,* P < 0.05).
tween percent penetration and LIN or B/CF did not exhibit this trend. The correlations between penetration index and CASA-generated semen parameters showed the same pattern as for percent penetration. The second difference between BWW and TYB treatment was that ALH was significantly correlated with SPA performance for samples treated with TYB but not those treated with BWW.
SPA Performance Groups-t Tests Velocity, beat/cross frequency, motility, concentration, and linearity were all significantly higher for penetrators (> 0% penetration, n = 76) than for nonpenetrators (0% penetration, n = 55) in the BWW-treated SPA (group 1, Fig. 2). Similarly, VEL, ALH, B E F , MOT, and CONC were significantly higher for penetrators (n = 119) than for nonpenetrators (n = 21) in the TYB-treated SPA (group 2). In group 3, VEL, ALH, BC/F, MOT, and CONC were significantly higher for high penetrators ( 2 20% penetration, n = 74) than for low penetrators (< 20% penetration, n = 66). Differences between penetration categories within groups were substantial for several motility parameters (e.g., group 2, 69.6 vs. 53.3 pm/sec for velocity) with the most pronounced differences in group 2 (Fig. 2). Nevertheless, the variation around the mean was high and the range of values overlapped considerably. The most notable difference between BWW- and TYBtreated SPA performance was that mean ALH was significantly different between penetration categories with TYB but not with BWW. Patients who penetrated
better in the SPA had higher ALH in their fresh samples.
SPA Performance Groups-Discriminant Function Analysis DFA correctly classified the SPA performance of 67% (51 out of 76) of the penetrators and 71% (39 out of 55) of the nonpenetrators from the CASA motility parameters of patients treated with BWW (group 1). For patients treated with TYB (group 2), DFA correctly classified the SPA performance of 76% (90 out of 119) of the penetrators and 86% (18 out of 21) of the nonpenetrators from their CASA motility parameters. The level of penetration for group 3 was similar to group 1 with DFA correctly classifying the SPA performance of 68% (53 out of 78) of the high penetrators and 68% (47 out of 69) of the low penetrators. For all groups, DFA classified significantly more SPA results correctly than if the result had been chosen by chance. However, percent correct classification was relatively high only for group 2. Motion parameters added significant information to the discriminant analysis. For example, in TYB with 0% penetration a s cutoff (group 2) inclusion of volume, concentration, and percent motility alone in the discriminant analysis produced only 67% correct classification compared to 81% when motion parameters were included. Also in this group, only 67% of nonpenetrators were correctly predicted with conventional semen parameters whereas 86% were correctly predicted when motion parameters were included in the DFA.
330
P.M. FETTEROLF AND B.J. ROGERS
Using CASA-generated motility parameters for group 2 patients, DFA predicted that 93 men would penetrate in the SPA with TYB. Of these patients, 90 (97%) successfully penetrated a t least one egg. The three men who failed to penetrate initially were able to penetrate in a subsequent SPA test or with BWW. Thus, the discriminant function equation can be used to accurately predict TYB-treated SPA penetrators (> 0%) from computerized motility parameters.
was comparably successful at classifying the SPA performance of patients based on motility parameters. In the best case (TYB-treated SPA, 0% vs. > 0% penetration), DFA correctly classified 76% of the penetrators and 86% of the nonpenetrators. Such levels of classification would seem to preclude the use of the DFA and motility parameters to replace the SPA. However, the discriminant function equation predicted that 93 patients would penetrate with TYB treatment and 90 (97%)succeeded. Of the three men who failed, two penetrated in subsequent SPAs and the other penetrated in the BWW-treated SPA. Thus, this discriminant equation can be used to accurately predict future successful penetrators from computerized motility characteristics. This discriminant function could have saved 63% (93 out of 147) of our patients the expense of the SPA, if a penetration threshold of 0% for the TYBtreated SPA is considered diagnostic in the infertility workup. It is concluded that objective motility parameters can predict hamster egg penetration (> 0%) in the TYBtreated SPA using multivariate statistical analysis (DFA). This approach appears to be a potentially useful, rapid alternative to the SPA for predicting sperm penetrating ability in a large subset of infertility patients.
DISCUSSION Some investigators have reported significant relationships between objectively determined motility parameters and SPA performance (Aitken et al., 1985; Holt et al., 1985; Ginsberg et al., 1987) but others have found no relationship (Kossoy et al., 1987). Results reported in this study support the hypothesis that the vigor of sperm motion is positively related to quantitative measures of penetration of zona-free hamster eggs. Measures of sperm motion such as velocity, beat-cross frequency and percent motility were all highly significantly (P < 0.001) related to penetration. In addition, these same parameters were significantly higher (P < 0.001) for relatively high penetrators than for low or unsuccessful penetrators when compared within the SPA performance groups. Unfortunately, the correlations and penetration-related differences in motility ACKNOWLEDGMENTS parameters reported herein are not sufficiently strong to permit prediction of SPA outcome with confidence. We thank Tom Thompson and Kay Gibson for perPrevious investigations that have found no relation- forming the SPAs and Kathy Carter and Charlene ship between computerized motility parameters and Thompson for help in preparation of the manuscript. SPA performance may have been confounded by lack of dilution of the semen sample (Critser et al., 1988; REFERENCES Vantman et al., 1988) or inappropriate settings for the Aitken RJ,Best FSM, Richardson DW, Djahanbakch D, Mortimer D, Templeton AA, Lees MM (1982): An analysis of sperm function in computer software (Knuth et al., 1987; Mack e t al., cases of unexplained infertility: Conventional criteria, movement 1988). It has been demonstrated that such factors can characteristics, and fertilizing capacity. Fertil Steril 38:212-2 19. greatly affect computerized motility analysis (Critser Critser JK, Colvin KE, Critser ES (1988): Effect of sperm concentraet al., 1988; Vantman et al., 1988; Knuth et al., 1987; tion on computer assisted semen analysis results. Andrology 9:14Mack et al., 1988). Using the same computer system a s P. in this study, Kossoy et al. (1987) found no strong re- Falk RM, Silverberg KM, Fetterolf PM, Kirchner FK, Rogers BJ (1990):Establishment of TEST-yolk buffer enhanced SPA limits for lationships between motility parameters and SPA perfertile males. Fertil Steril 54:121-126. formance. In that study, semen samples were not di- Ginsberg KA, Moghissi KS, Abel EL, Ager JW (1987): Predicting luted and computer settings were not optimal (Knuth hamster egg penetration by computer-assisted sperm movement analysis. Fertil Steril 48:40. et al., 1987; Mack et al., 1988). In comparison, samples in this study were diluted to a concentration within the Holt WV, Moore HDM, Hillier SG (1985): Computer assisted measurement of sperm swimming speed in human semen: Correlation optimal range for CellSoft (Vantman et al., 1988) and of results with in vitro fertilization assays. Fertil Steril44:112-119. the computer settings were very similar to those which Hope E, Blackburn K, Zenick H, Rafales L, Meyer C (1984): Computhave been determined to be optimal (Mack et al., 1988). erized evaluation of human sperm. Biol Reprod 30:176. In a n unpublished study similar to this one, Gins- Karp LE, Williamson RA, Moore De, Shy KK, Plymate R, Smith WD (1981): Sperm penetration assay: Useful test in evaluation of male berg and colleagues (1987) compared computer-generfertility. Obstet Gynecol 57:620-623. ated motility parameters with percent penetration in Knuth UA, Yeung C, Nieschlag E (1987): Computerized semen analthe BWW-treated SPA. They divided patients into 0ysis: Objective measurement of semen characteristics is biased by 10% and > 10% penetration groups and reported that subjective parameter setting. Fertil Steril 48:118-124. discriminant function analysis correctly classified 81% Kossoy LR, Thompson T, Buck A, Hinson M, Brodie B, Vaughn W, Wentz AC, Rogers BJ (1987): Comparison of the sperm penetration of the samples in the low penetration group and 60% of assay (SPA) with objective motility parameters. Andrology 8:38-P. samples in the high penetration group. In our study Mack SO, Wolf DP, Tash JS (1988): Quantitation of specific paramewith a larger sample, different SPA methodologies, and ters of motility in large numbers of human sperm by digital image processing. Biol Reprod 38:270-281. different group cutoffs, discriminant function analysis
SPA AND MOTILITY PARAMETERS Mathur S, Carlton M, Ziegler J , Rust PF, Williamson HO (1986): A computerized sperm motion analysis. Fertil Steril 46:484-488. Milligan MP, Harris S, Dennis KJ (1980): Comparison of sperm velocity in fertile and infertile groups as measured by time lapse photography. Fertil Steril 34:509-511. Overstreet JW, Yanagamachi R, Katz DF, Hayashi K, Hanson FW (1980): Penetration of human spermatozoa into the human zona pellucida and the zona-free hamster egg: A study of fertile donors and infertile patients. Fertil Steril 33534-542. Rogers BJ (1983): Hamster egg: evaluation of human sperm using in vitro fertilization. In PG Crosignai (ed): “In Vitro Fertilization and Embryo Transfer, Proceedings of the Serono Symposium.” New York: Academic Press, p 101. Rogers BJ (1985):The sperm penetration assay: Its usefulness reevaluated. Fertil Steril 43:821-840. Rogers BJ (1986): The usefulness of the sperm penetration assay in predicting in vitro fertilization (IVF) success. J Vitro Fertil Embryo Transfer 3:209 -2 11. Rogers BJ, Holmgren W, Krivacka T, Maxson WS, Wentz AC (1985a):
331
Enhancement of fertilizing capacity by TEST-yolk buffer. 41st Annual Meeting of the American Fertility Society, Chicago, Illinois. The Program Supplement, p. 46, number 83. Rogers BJ, Mygatt GG, Soderdahl DW, Hale RW (1985b): Monitoring of suspected infertile men with varicocele by the sperm penetration assay. Fertil Steril 44:800-805. Rogers BJ, Van Campen H, Ueno M, Lambert H, Bronson R, Hale R (1979): Analysis of human spermatozoa1 fertilizing ability using zona-free ova. Fertil Steril 32:664-670. Smith RG, Johnson A, Lamb D, Lipshultz LI (1987): Functional tests of spermatozoa. Urol Clin N Am 14:451-458. Vantman D, Koukoulis G, Dennison L, Zinaman M, Sherins R J (1988): Computer-assisted semen analysis: Evaluation of method and assessment of the influence of sperm concentration on linear velocity determination. Fertil Steril 49:510-515. Yanagamachi R, Yanagamachi H, Rogers BJ (1976): The use of zonafree animal ova as a test system for the assessment of the fertilizing capacity of human spermatozoa. Biol Reprod 15:471-476.
