Nabil Aziz, M.R.C.O.G.,* Simon Fear, B.Sc., Clare Taylor, Ph.D., Charles R. Kingsland, M.D., and D. Iwan Lewis-Jones, M.D.*

Transcription

1 FERTILITY AND STERILITY VOL. 70, NO. 5, NOVEMBER 1998 Copyright 1998 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A. REPRODUCTIVE BIOLOGY Human sperm head morphometric distribution and its influence on human fertility Nabil Aziz, M.R.C.O.G.,* Simon Fear, B.Sc., Clare Taylor, Ph.D., Charles R. Kingsland, M.D., and D. Iwan Lewis-Jones, M.D.* University of Liverpool and Liverpool Women s Hospital, Liverpool, United Kingdom Received January 7, 1998; revised and accepted July 20, Supported in part by grant RDS 1720, Research Development Fund, University of Liverpool, Liverpool, United Kingdom. Reprint requests: Nabil Aziz, M.R.C.O.G., Liverpool Women s Hospital, Crown Street, Liverpool L8 7SS, United Kingdom. * Academic Department of Obstetrics and Gynaecology, University of Liverpool. Department of Mathematical Sciences, Division of Statistics and Operational Research. University of Liverpool. Reproductive Medicine Unit, Liverpool Women s Hospital /98/$19.00 PII S (98) Objective: To study the distribution of live sperm head size in semen and sperm preparations as a predictor of fertility. Design: Prospective blind clinical trial. Setting: Academic tertiary referral center. Patient(s): One hundred fifty-five patients undergoing IVF treatment. Females with conditions negatively influencing fertilization were excluded. Intervention(s): Morphometric analysis (head area, major axis, minor axis, and elongation ratio) of video images of sperm in semen and swim-up preparations used for IVF treatment was performed with a Hamilton-Thorne analyzer V 8.1 (Hamilton-Thorn Research, Beverly, MA). Main Outcome Measure(s): Oocyte fertilization. Result(s): Seventy-four percent of patients achieved fertilization. Fertilizers and nonfertilizers had different sperm head area distribution. The fertilizers had a significantly smaller interquartile range of sperm head area and of major axis in both semen and sperm preparation compared with the nonfertilizers. A subgroup of men who had fathered a child naturally had a more uniform sperm head area in semen with a significantly smaller median compared with those who failed to father a child naturally with their healthy female partner. We used multiple logistic regression applying forward stepwise selection of variables in building three predictive models of probability of fertilization. Conclusion(s): Successful IVF or history of fathering a child was associated with a more uniform sperm head area in semen and sperm preparation. (Fertil Steril 1998;70: by American Society for Reproductive Medicine.) Key Words: Sperm, live morphometry, sperm deformity index, in vitro fertilization, human Evaluation of male fertility is based predominantly on the results of sperm count, sperm motility, and sperm morphology. Sperm count and motility are imperfect discriminators largely because of the great overlap between fertile and nonfertile samples in both in vivo (1) and in vitro (2, 3) fertility studies. On the other hand, sperm morphology assessment is riddled with subjectivity, although the sophistication brought into sperm morphological examination by the recent inclusion of the stricter morphological criteria (4) and sperm deformity index (3) has enhanced this parameter s predictive power and reproducibility. The study of fixed sperm head morphometric characteristics using photomicrography was first attempted in the bull (5) and then in the human (6). Moench and Holt (6) observed that semen samples of infertile men often exhibited both a high percentage of abnormal sperm heads and a high coefficient of variation of head size ( 11%) and concluded that adequate semen analysis should assess sperm count, motility, morphology, and head biometry. One of the disadvantages of studying sperm head size in a fixed stained sperm is the lack of concurrent knowledge of sperm viability or motility. Moreover, sperm fixation and staining result in shrinkage of the sperm head with a reduction in its length, width, and projected surface area by 14%, 16%, and 30%, respectively (7). The extent of this size reduction depends on the staining technique used, giving rise to different normal ranges of head dimensions for different stains (8). Thus, the need to concentrate on obtaining biologic data on, and measurements of, spermatozoa that are functionally active remains a 883

2 priority (9). This is addressed in part by selecting the motile subpopulation in a semen sample before studying its morphometric feature in the dried fixed smear (10). Although the use of videomicrography and the automation of image analysis has made live sperm head morphometry more feasible (11, 12), inherent problems in this technique (e.g., sperm collision) can compromise the validity of its results. The correlation between male fertility and head size distribution of live sperm, using this new technique, was reported in only two studies in relation to the achievement of pregnancy both naturally (13) and in the context of a donor insemination program (14). We used human IVF as a suitable model to study prospectively the influence of live sperm head morphometric distribution on sperm-egg interaction. We also investigated a possible correlation between live sperm automated morphometry and morphology as assessed in fixed stained smears applying strict criteria. MATERIALS AND METHODS Couples undergoing IVF treatment in the Reproductive Medicine Unit of Liverpool Women s Hospital between April 1993 and June 1994 were considered for this study. Approval for the project was obtained from the local ethics committee. Female partners who were 40 years old or those with a history of severe endometriosis or polycystic ovary syndrome were excluded because these could independently affect fertilization rates. Treatment cycles that resulted in an unfavorable ovarian response ( 2 mature oocytes collected) also led to exclusion from the study. All male partners who were not azoospermic were included in the study irrespective of their semen quality to avoid biased interference with the dynamic ranges of the features tested. Ovarian stimulation was standard and incorporated the GnRH analogue long protocol and hmg. The standard Reproductive Medicine Unit protocols for follicular monitoring, oocyte recovery, and oocytes and embryo culture have been described elsewhere (15). Insemination was performed 40 hours after hcg with use of 100,000 prepared motile sperm per oocyte. At hours after hcg administration, oocytes were examined for evidence of normal fertilization by observing the presence of two pronuclei. Tripronucleate oocytes were noted and included in the total number of fertilized oocytes. Standard Semen Analysis and Preparation Semen samples were obtained by masturbation on the morning of egg retrieval after 4 days of sexual abstinence and allowed to liquefy for 30 minutes at room temperature. Semen volume and sperm count and motility were measured according to World Health Organization (WHO) criteria (16). Semen samples were prepared for insemination by a standard swim-up technique as described elsewhere (15). Thin smears of semen and sperm preparations stained with Spermac stain (Stain Interprise, Wellington, Republic of South Africa) were used to assess sperm normal morphology by applying strict criteria (4). The sperm deformity index also was assessed in both states as described elsewhere (3). Briefly, the sperm deformity index is calculated by dividing the total number of deformities observed by the number of sperm randomly selected and evaluated irrespective of their morphological normality. We demonstrated that a sperm deformity index of 1.6 was associated with poor fertilization (3). Video Recording of Semen and Sperm Preparation Samples Images for all semen and swim-up sperm preparations generated by the CCD video camera of Hamilton-Thorne analyzer (Hamilton-Thorne Research, Beverly, MA) were recorded with a video camera. These recordings were used later for morphometric analysis. An S-VHS multisystem videocassette recorder (VCR; Panasonic NV-W1E, Japan) and double-coated videotapes (Fuji, Super XG PRO; Fuji, Tokyo, Japan) suitable for master recording were used to offer sharper video images. A high focal-depth 10 negative-phase contrast objective in the optical system produced an image of the specimen. To reduce the incidence of sperm collision and standardize system settings, a semen sample with counts of /ml was diluted with patient s own seminal plasma obtained by centrifugation (600 g for 10 minutes) of the remaining semen sample. Sperm preparations used for the IVF treatment were sampled before dilution to ensure an adequate number of sperms in the captured fields. Each footage did not exceed 2 minutes to avoid sample dryness in the Makler chamber (Sefi Medical Instruments, Haifa, Israel). Images from a minimum of 10 microscopic fields were included in each footage to ensure reliable sampling. Peripheral fields were avoided because of the distorted images of sperm heads encountered there. Care was taken to keep sperm in focus when moving to a new field. Sperm Head Morphometry Sperm head morphometry was analyzed with use of the same recording VCR and the same analyzer settings for brightness and image magnification. The video-recorded image was sampled for 1 second at an acquisition rate of 30 Hz (giving 30 frames) and digitally encoded in a pixel array into digital memory space. Each pixel has one of 64 levels of light intensity and corresponds to an area of 1.7 m 2 of sperm head. The surface area in square micrometers, the length (major axis) and the width (minor axis) in micrometers, and the elongation ratio (width/length) for each motile sperm head image are calculated in the series of frames in which the sperm is captured in the field. The minor axis algorithm finds the maximum thickness in direction normal to the major axis. Symmetry is assumed in the calculation. 884 Aziz et al. Live Sperm head morphometry Vol. 70, No. 5, November 1998

3 TABLE 1 Comparison of patient characteristics and sperm count and motility between the fertilizers and nonfertilizers. Criterion Fertilizers (n 114) Nonfertilizers (n 41) 95% CI for difference between medians (two-sided P) Median female age (range) (y) 34 (24 39) 34 (25 39) 1 to2(p.38) Median period of infertility (range) (y) 6 (1 17) 7 (1 16) 1 to1(p.57) Median no. of oocytes inseminated (range) 5 (2 22) 5 (2 13) 1 to1(p.95) Median sperm count (range) ( 10 6 /ml) 95.5 (9 560) 46 (10 301) 66 to 29 (P.0001) Median sperm motility (range) (%) 72 (20 98) 59 (4 93) 24 to 10 (P.0001) The area of the sperm head is calculated by counting the number of pixels contained within the edges of its digitized image. The analyzer designates the largest image of the sperm head during the acquisition time as the optimum sperm for morphometry on the assumption that it represents the head at the moment where it is lying flat. In this study when the largest area was due to momentary collision with another sperm, an alternative largest image was selected by the investigator (N.A.). A total of 200 sperm was assessed for each semen and sperm preparation sample. Quality Control Initially, the analyzer measurement scale was validated when objects were lying at different angles to the horizontal. The analyzer internal quality control function also was used to determine the acquisition accuracy and the appropriateness of gate settings. This ensured that only sperm (not debris) were recognized in a field and that static objects and motile objects were genuinely so. Statistical Methods Data were analyzed with inbuilt functions within the Statistical Package for Social Science (SPSS UK Ltd, Chertsey, Surrey, United Kingdom). For all morphometric measurements we used median, interquartile range, and range as summary statistics. Nonparametric tests were used for examining significant morphometric differences between ejaculate and swim-up sperm preparations and/or between subgroups, taking into account the ejaculate-sperm preparation pairing whenever applicable. We also performed Student s t-tests (paired as applicable) whenever we found approximately symmetric sample distributions; no such tests differed in the strength of their conclusions from the nonparametric tests reported here. We used multiple logistic regression applying forward stepwise selection of morphometric variables, based on likelihood ratio, in building predictive models of the probability of fertilization given sperm morphometry. RESULTS One hundred fifty-five couples underwent one treatment cycle each, of which 114 couples (73.5%) achieved fertilization and 41 couples (26.5%) did not. The oocyte fertilization rates varied between 10% and 100% (median, 87%). A complete set of morphometric measurements of sperm in semen and sperm preparation was available in 147 patients. Eight patients had morphometric data for semen only. The mean age of the female partner, the period of infertility, and the median number of mature oocytes inseminated was not significantly different in patients achieving fertilization and those who did not (Table 1). However, the sperm count and progressive motility in semen were significantly higher in the fertilizers than in the nonfertilizers (Table 1). Head Size Distribution of Ejaculated and Prepared Sperm The distribution of the sperm head size in semen and swim-up sperm preparations were compared for the 147 patients who had complete sets of data. The location of the distribution of sperm head area of the sperm population in the swim-up preparation displayed a shift to the right compared with that for the ejaculated sperm (Fig. 1). The medians of sperm head area, major axis, and minor axis were significantly larger in the sperm preparations than in ejaculated semen (Table 2). The median ranges of sperm head major axis, minor axis, and elongation were significantly smaller in the sperm preparation compared with semen (Table 2). Morphometry and Outcome of Fertilization The median, range, and the interquartile range of the four morphometric parameters of the sperm head in the fertilizers and nonfertilizers were compared (Tables 3 and 4). Spermatozoa in semen and swim-up preparation in the fertilizers had a significantly smaller interquartile ranges of head area and head major axis than the nonfertilizing group (Fig. 2). The shift in the location of the sperm head area distribution as a result of the swim-up selection of spermatozoa was similar in both outcome groups and mirrored that seen for the whole population (see above). Logistic Regression Models Logistic regression was used to evaluate the usefulness of morphometric data alone in predicting the achievement of IVF. First, the morphometric data for seminal sperm were FERTILITY & STERILITY 885

4 FIGURE 1 Distribution of sperm head area in semen and sperm preparation. analyzed with use of forward stepwise selection. Only the range and the interquartile range of sperm head area were selected (model one) (Table 5). Next, the morphometric data of sperm preparation were offered alone for stepwise selection, and only the sperm head area interquartile range and the range of elongation were selected (model 2). Last, when the complete set of data of morphometry of sperm in semen and sperm preparation were available for selection, the only three TABLE 2 Difference in morphometric features between sperm in the ejaculated semen and the corresponding swim-up preparation using Wilcoxon s signed rank test. Parameter Ejaculated semen Sperm preparation Median Range Median Range 95% CI for difference between medians Two-sided P Area ( m 2 ) Head area to 3.8 P.0001 Range of head area to 1.4 P.88 Interquartile range of head area to.4 P.69 Major axis ( m) Major axis to.53 P.0001 Range of major axis to 2 P.0001 Interquartile range of major axis to.2 P.01 Minor axis ( m) Minor axis to.4 P.0001 Range of minor axis to.4 P.005 Interquartile range of minor axis to.04 P.3 Elongation Elongation to.005 P.3 Range of elongation to.1 P.0001 Interquaratile range of elongation to.02 P Aziz et al. Live Sperm head morphometry Vol. 70, No. 5, November 1998

5 TABLE 3 Morphometric features of sperm in ejaculated semen of patients who achieved fertilization (n 114) and those who did not (n 41). Median Median interquartile range Median range Parameter Fertilizers Nonfertilizers Fertilizers Nonfertilizers Fertilizers Nonfertilizers Head area ( m 2 ) * 7* Major axis ( m) Minor axis ( m) Elongation * P.023 (median difference 0.6; confidence interval ). P.04 (median difference.11; confidence interval 0.3). terms that were significant were sperm head area interquartile range and range of elongation in sperm preparation, and the sperm head area range in semen (model 3). To study the fit of these models in terms of predictive quality, we gathered the fitted probability of fertilization into three bands (.5,.5.85,.85 or broadly unlikely, likely, and very likely to fertilize). Model 2 provided the best fit especially for the nonfertilizers. In the.5.85 band the three models provided similar prediction with 74% of patients within this band being correctly predicted as fertilizers. Model 1 featured 82% of all patients (127 of 155) within this probability band. However, approximately 74% fertilization is the overall average, thus model 1 has in general little to offer other than identifying a small group as highly likely to fertilize. Models 2 and 3 discriminate much more, although model 3 greatly overpredicts nonfertilization. Morphology and Morphometry The relationship between automated morphometric measurements of motile sperm and the strict morphological assessment of sperm in fixed stained smears obtained from the same samples was examined. The proportion of normal sperm morphology correlated negatively with the interquartile range of sperm head area (semen: Kendall s tau b.27, P.0008; swim-up: Kendall s tau b.27, P.0001). On the other hand, there was a significant positive correlation between the sperm deformity index and the interquartile range of sperm head area (semen: Kendall s tau b.12, P.02; swim-up: Kendall s tau b.19, P.001). The proportion of large sperm head or small sperm head in semen and in sperm preparation did not significantly correlate to the median, interquartile range, or the range of sperm head area. The proportion of tapered head correlated negatively to the median elongation and positively with the interquartile range of sperm head elongation in sperm preparations (Kendall s tau b.18, P.0019; and Kendall s tau b.13, P.02, respectively) but not in ejaculated semen. Sperm Morphometry and Natural Fertility We identified 24 men among our study population who gave a history of fathering one or more children naturally in current or previous relationships (subgroup 1). Their current infertile relations were attributed only to tubal damage. Another 41 men had healthy female partners but never fathered a child naturally in current or previous relations (subgroup 2). Although the sperm count and motility were similar in the two subgroups, subgroup 1 had significantly higher normal TABLE 4 Morphometric features of sperm in the sperm preparations of patients who achieved fertilization (n 109) and those who did not (n 38). Median Median interquartile range Median range Parameter Fertilizers Nonfertilizers Fertilizers Nonfertilizers Fertilizers Nonfertlizers Head area ( m 2 ) * 7.2* Major axis ( m) Minor axis ( m) Elongation * P.0001 (median difference.9; confidence interval.7 1.8). P.007 (median difference.13; confidence interval.05.3). FERTILITY & STERILITY 887

6 FIGURE 2 Distribution of sperm head area in sperm preparations in the fertilizers and nonfertilizers. sperm morphology and a significantly lower sperm deformity index than subgroup 2 (Table 6). The median, the interquartile range, and the range of the sperm head area were significantly smaller in subgroup 1 than subgroup 2 (Table 6). However, the major axis, minor axis, and elongation were not significantly different in the two groups. DISCUSSION TABLE 5 The proportion of patients achieving fertilization in a predictive band using three logistic regression models of morphometric parameters of semen and/or sperm preparation. P value Ejaculated semen (%)* Sperm preparation (%) Ejaculated semen and sperm preparation (%).5 4/10 (40) 5/19 (26) 13/23 (57) /127 (72) 51/69 (74) 50/66 (75).85 17/18 (94) 53/59 (90) 71/89 (81) * Model using sperm head area range and interquartile range. Model using sperm head area interquartile range and range of elongation. Model using sperm head area interquartile range and range of elongation in sperm preparation and sperm head area range in semen. To our knowledge this is the first study to describe the live sperm head morphometric distribution in patients undergoing standard IVF using the outcome of sperm-oocyte interaction as a biologic endpoint. The automated image analysis module in the Hamilton-Thorne computer-aided semen analysis system V 8.1 was used to show for the first time that the swim-up sperm population had greater uniformity in head dimensions and significantly larger medians of head area, major axis, and minor axis than ejaculates. The median values of the four morphometric features of the sperm head in semen and sperm preparation were inefficient in discriminating between the two outcome groups. There was, however, a significant difference in sperm head size distribution between these two groups, with the fertilizers displaying a greater uniformity (smaller interquartile ranges) in sperm head area and major axis in both semen and the swim-up preparation than the nonfertilizers. There was no evidence that the outcome of sperm-egg interaction was associated with any special shift in the distribution of sperm head size in the swim-up preparation compared with the ejaculated semen. All three logistic models we explored selected only attributes of distribution (interquartile ranges and ranges) and rejected population central values. Model 2 provided the best fit, especially for nonfertilizers. Accordingly, we offer model 2 as the predictive model of choice, but we emphasize that this is based solely on sperm morphometry and ignores other known variables such as sperm count and motility. Sperm head area interquartile range in sperm preparation is a significant predictor even in a model including sperm motility (Aziz et al., manuscript in preparation). 888 Aziz et al. Live Sperm head morphometry Vol. 70, No. 5, November 1998

7 TABLE 6 Ejaculated semen sperm count, motility, and head area in naturally fertile men (subgroup 1) and subfertile men (subgroup 2). Sperm parameter Subgroup 1 (n 24) Subgroup 2 (n 41) 95% CI for difference between medians (two sided P) Median sperm count (range) ( 10 6 /ml) 56 (10 301) 75 (22 208) 42 to 11 (P.35) Median motility (range) (%) 67 (40 90) 60 (4 96) 15 to 5 (P.29) Median normal sperm morphology (range) (%) 11 (4 23) 5 (0 26) 1 9 (P.01) Median sperm deformity index (range) 1.41 ( ) 1.54 ( ) (P.018) Median head area ( 2 ) (P 0.005) Median interquartile range of head area ( 2 ) (P.013) Median range of head area ( 2 ) (P.01) Among our study population, the 24 men who had fathered a child naturally had a significantly smaller median sperm head area than that for the 41 men who had failed to father a child naturally with their healthy partners. The naturally fertile group also had a significantly smaller range and interquartile range of sperm head area, indicating a greater uniformity of sperm head size in semen than the other subgroup. The other morphometric variables studied and the sperm count and motility in semen did not differ significantly between the groups. Irvine et al. (14) observed a similar statistically significant difference in sperm head area when they examined the predictive value of live sperm head morphometry for the achievement of pregnancy in vivo using the Hamilton- Thorne HTM-S However, the mean sperm head area they obtained for those who conceived was 24.5 m 2 and for those who remained infertile for the follow up period was 26.8 m 2. These figures are almost 40% larger than those we observed (Table 6). Irvine et al. (14) also reported that the major and minor axes were different in the two group of patients, which we did not observe. Although these 24 naturally fertile men represent a small group, it is notable that they had the smallest sperm head area in the whole study population. This may indicate that, when it comes to natural fertility, a different set of rules apply compared with those involved in the outcome of oocyte fertilization in vitro as a biologic endpoint. An obvious pitfall in live sperm morphometry is the inability of the analyzer to recognize that the largest sperm head area it selected may be due to sperm collision or fuzziness of the sperm image. The incidence of sperm col- FIGURE 3 Distribution of sperm head area in the semen of natural fathers and men in unexplained infertile relationships. FERTILITY & STERILITY 889

8 lision was reduced in this study by diluting semen samples using homologous seminal plasma when sperm count exceeded /ml. Sperm collision was not a prominent feature in swim-up preparations. On the other hand, sperm moving out of focus was minimized with use of the Makler chamber with relatively smaller depth. The reliability of the morphometric data was enhanced by visually choosing the alternative largest image if the optimum image selected by the analyzer was deemed inappropriate due to collision or fuzziness. The summary statistics offered by the analyzer at the end of each field analysis was rejected because it was exaggerated by the erroneous data arising from sperm collision. Moreover, the analyzer calculated the mean and SD for data that did not display normal distribution. The analogue image produced by the analyzer camera is digitized in a pixel array with a 64-gray scale gradation for intensity. The brightness of each pixel in the image of the head is proportional to its phase shift, and therefore is closely proportional to the cell mass traversed by the light beam ending on that pixel. The area of the sperm head is calculated by counting the number of pixels contained within the edges of its digitized image. The edge of the sperm head is defined by the pixels exceeding the combined mean background intensity and minimum contrast intensity. The latter is measured in arbitrary units of brightness and in this study was set at 8. Thus, the apparent size of the sperm head depends on the illumination level: higher illumination levels increases the radial distance from the center of mass to the minimum brightness (contrast) contour, which defines the head edge. Therefore, sperm head looks larger at higher illumination. The analyzer illumination intensity was identical for all the samples analyzed in this study to maintain the relationship of the relative sizes of sperm heads giving valid measurements. The pixel is the smallest unit in image analysis. The area of sperm head represented by a single pixel depends on image magnification and resolution. High magnifications using oil immersion lens are possible when studying static sperm such as in fixed stained smear. This gives higher resolution, and the pixel size corresponds to m of head size. Tracing motile images necessitates lower magnification to enable the transfer of up to 60 images per second (30 images/s in this study) to enable the sorting out of truly motile sperm from nonmotile sperm or those showing gosling movement. Under these circumstances, the image resolution is less and the pixel size is larger, and in our study it corresponds to 1.7 m 2 of sperm head. Although the measurement may appear to be crude, yet, unlike static sperm, the size is determined in real time with full knowledge of sperm motility status and without altering the real dimensions as a result of sperm fixation and staining. Katz et al. (7) studied immobilized live spermatozoa in men with proven fertility with use of phase contrast optics and a 100 oil immersion lens with a final magnification at the point of measurement of They reported the mean major and minor axes ( 2 SD) as 5.26 m ( ) and 3.37 m ( ), respectively. With use of the Hamilton-Thorne analyzer, Irvine et al. (14) calculated the median major and minor axes of live sperm head in the ejaculates of similar group of patients who conceived naturally at 8.3 m (interquartiles ) and 4.4 m (interquartiles ), respectively. They argued that the discrepancy with the estimates of Katz et al. resulted from the difference in the magnification of the image measured. The median (and interquartiles) of the sperm head major and minor axes among the subgroup of men in our study population who had fathered a child naturally was calculated at 6.2 m ( ) and 3.9 m ( ), respectively. These measurements obtained on a smaller image magnification are comparable with those reported by Katz et al. (7) with use of high magnification but are significantly different from those reported by Irvine et al. (14) with use of Hamilton-Thorne analyzer at low magnification. The discrepancy in measurements made with the Hamilton-Thorne analyzer in the two studies may, in part, be due to the use of dissimilar settings and gates of the analyzer. More significantly, however, Irvine et al., unlike our study design, accepted the analyzer s summary data that were inflated by the inclusion of large sperm heads resulting from collision. In view of the obvious similarities between our measurements and those made by Katz et al. (7), we believe that the calculations based on the smaller image magnifications of the Hamilton-Thorne analyzer are realistic and can be of clinical value. When the morphometric measurements were correlated with the strict morphology of stained smears of the same sample, we found that the larger the proportion of morphologically normal sperm or the smaller the sperm deformity index the greater the uniformity of head area in the sperm population both in semen and sperm preparation. There was also a significant negative interdependency between the proportion of the tapered head form in the swim-up and the median elongation ratio. Macleod et al. (17) failed to observe any correlation between live sperm morphometry, using the Hamilton- Thorne analyzer HMT 2030, and sperm morphology in semen applying WHO criteria of Although it seems logical to assume a correlation between morphology and morphometry of sperm head, objectivity and accuracy in assessing these two parameters are required to reflect this mutual interdependency. In conclusion, this is the first study on live sperm to demonstrate that successful oocyte fertilization is associated with a sperm population with greater uniformity of head size 890 Aziz et al. Live Sperm head morphometry Vol. 70, No. 5, November 1998

9 both in semen and the sperm preparation. This observation was also true for the group of men in our study population who had fathered a child naturally. This subgroup of men had a significantly smaller sperm head compared with the rest of the population. The Hamilton-Thorne analyzer made live sperm morphometry attainable within a clinical setting with reasonable speed and convenience. It calculated the live sperm head dimensions that were comparable with previously reported values, which were obtained by using complicated and time-consuming techniques that provided highly magnified images. We were able to demonstrate mutual interdependency between morphological normality and the uniformity of head size in a sperm population. However, we believe we were able to achieve this degree of consistency in our results by following a strict methodology to avoid potential pitfalls inherent in image analysis of motile objects. We also believe that the morphometric parameters should be considered in combination with other variables to provide a stronger prediction of potential fertility, and this is currently under investigation. References 1. Plansky FF, Lamb EJ. Do the results of semen analysis predict future fertility? A survival analysis study. Fertil Steril 1988;49: Edwards RG, Fishel SB, Cohen J, Fehilly CB, Purdy JM, Slater JM, et al. Factors influencing the success of in vitro fertilization for alleviating human infertility. J In Vitro Fert Embryo Transf 1984;1: Aziz N, Buchan I, Taylor CT, Kingsland CR, Lewis-Jones I. The sperm deformity index: a reliable predictor of the outcome of oocyte fertilization in vitro. Fertil Steril 1996;66: Kruger TF, Menkveld R, Stander FSH, Lombard CJ, Van der Merwe JP, Van Zyl JA. Sperm morphologic features as a prognostic factor in in vitro fertilization. Fertil Steril 1986;46: Williams WW, Savage A. Observations on the seminal morphology of bull. Cornell Vet 1925;15: Moench GL, Holt H. Biometrical studies of head lengths of human spermatozoa. J Lab Clin Med 1932;17: Katz DF, Overstreet JW, Samuel SJ, Niswander PW, Bloom TD, Lewis EL. Morphometric analysis of spermatozoa in the assessment of human male fertility. J Androl 1986;7: Menkveld R, Oettle EE, Kruger TF, Swanson RJ, Acosta AA, Oehninger S. Atlas of human sperm morphology. Baltimore: Williams & Wilkins, 1991: Barratt CLR, Naeeni M, Clement S, Cooke ID. Clinical value of sperm morphology for in vivo fertility: comparison between World Health Organization criteria of 1987 and Hum Reprod 1995;10: Mundy AJ, Ryder TA, Edmonds DK. Morphometric characteristics of motile spermatozoa in subfertile men with an excess of non-sperm cells in the ejaculate. Hum Reprod 1994;9: Katz DF, Overstreet JW. Sperm motility assessment by videomicrography. Fertil Steril 1981;35: Schoevaert D. Automated recognition and morphological analysis of human spermatozoa. In: Robard D, Forti G, editors Computers in endocrinology. New York: Raven Press, 1984: Macleod IC, Irvine DS. The predictive value of computer-assisted semen analysis in the context of a donor insemination programme. Hum Reprod 1995;10: Irvine DS, Macleod IC, Templeton AA, Masterton A, Taylor A. A prospective clinical study of the relationship between the computerassisted assessment of human semen quality and the achievement of pregnancy in vivo. Hum Reprod 1994;9: Kingsland CR, Aziz N, Taylor CT, Manasse PR, Haddad N, Richmond DH. Transport in vitro fertilization a novel scheme for communitybased treatment. Fertil Steril 1992;58: World Health Organization. Laboratory manual for the examination of human semen and sperm-cervical mucus interaction. 3rd ed. New York: Cambridge University Press, Macleod IC, Irvine DS, Masterton A, Taylor A, Templeton AA. Assessment of the computer-assisted image analysis: evaluation of the Hamilton-Thorne motility analyzer in the context of a service andrology laboratory. Hum Reprod 1994;9: FERTILITY & STERILITY 891