The sperm deformity index: a reliable predictor of the outcome of oocyte fertilization in vitro*

Transcription

1 FERTILITY AND Vol. 66, No.6, December 1996 Copyright 1996 American Society for Reproductive Medicine Printed on acid-free paper in U. S. A. The sperm deformity index: a reliable predictor of the outcome of oocyte fertilization in vitro* Nabil Aziz, M.R.C.O.G.t:!: lain Buchan, M.B., Ch.B. Clare Taylor, B.Sc.11 Charles R. Kingsland, M.D.II lwan Lewis-Jones, M.D.t University of Liverpool, and Liverpool Women's Hospital, Liverpool, United Kingdom Objective: To evaluate a novel expression of sperm morphological parameters, the sperm deformity index, as a predictor of fertilization in vitro. Design: Prospective blind clinical trial. Setting: Academic tertiary referral center. Intervention(s): Detailed sperm morphological assessment applying strict morphological criteria and a multiple entry technique for an unselected male population undergoing IVF. The sperm deformity index, defined as the average number of deformities per sperm assessed, was calculated. Patient(s): One hundred fifty-eight patients undergoing IVF treatment. Females with conditions negatively influencing fertilization were excluded. Main Outcome Measure(s): Fertilization rates and pregnancy. Result(s): Seventy-three percent of patients achieved fertilization. Patients achieving fertilization had a significantly higher median proportion of normal forms and a significantly lower median sperm deformity index than the nonfertilizers. The receiver operator characteristic (ROC) curves identified cutoff points that maximized the sum of sensitivity and specificity at sperm deformity index 1.6 and normal forms 4%. The sperm deformity index had a greater sensitivity (96%), specificity (72%), positive predictive value (90%), and negative predictive value (86%) than the proportion of normal sperm morphology (87%, 69%, 89%, 66%, respectively) at the optimal cutoff points. The area under the ROC curve was greater for the sperm deformity index (0.875) than for the proportion of normal sperm morphology (0.622). Achieving pregnancy did not correlate with sperm morphology. Conclusion(s): The sperm deformity index is a more reliable predictor of the outcome of fertilization in vitro than the proportion of normal sperm morphology and can assist to identify patients who require techniques such as intracytoplasmic sperm injection. Fertil Steril 1996;66: Key Words: Sperm, morphology, sperm deformity index, in vitro fertilization, human Received April 23, 1996; revised and accepted June 26, * Supported in part by grant RDS 1720, research development fund, University of Liverpool, Liverpool, United Kingdom. t Academic Department of Obstetrics and Gynaecology, University of Liverpool. :j: Reprint requests: Nabil Aziz, M.R.C.O.G., Liverpool Women's Hospital, Crown Street, Liverpool, L8 7SS, United Kingdom (FAX: +44 (0) ) Aziz et al. Sperm deformity index The goal of estimating correctly a man's fertility potential has long been of great interest to researchers. The advent of IVF has provided an ideal model with a defined outcome to assess the value of different semen parameters as predictors of the fertility potential of a semen sample without the influence of other fertility factors. In early studies, standard semen analysis, including sperm count, motility, and morphology proved to be a relatively poor predictor of the likely outcome of sperm-oocyte interaction, as 15% to 29% of cases showing normal semen characteristics were reported to have failed fertilization (1, Departments of Primary Care and Medicine, University of Liverpool. IIReproductive Medicine Unit, Liverpool Women's Hospital.

2 T 2). Sperm morphology as evaluated by the World Health Organization (WHO) criteria of 1987 (3) also was found to only have limited predictive value for oocyte fertilization in vitro (4-8). However, sperm morphology was found to correlate more closely with fertilization rates than sperm count and motility (7,8). Nonhuman sperm display more morphological uniformity than human sperm. At the turn of this century, this led to the recognition of the relationship between abnormal sperm forms and male infertility in the bull (9). In man, in whom morphologically normal and abnormal spermatozoa normally coexist in the same ejaculate, this link remains controversial. Conclusive evidence of the spontaneous fertilizing ability of a single human spermatozoon drawn from objective evaluation of both shape and size is technically unachievable. The correlation between the morphological pattern of spermatozoa in an ejaculate and its fertility potential is an alternative strategy. However, this only allows the probability analysis of the fertility potential of a particular semen sample and relies on strict reproducible techniques of assessing sperm morphology. Stricter criteria for normal sperm morphology were proposed as a more reliable predictor of the outcome of oocyte fertilization in vitro by Kruger and colleagues (10, 11). According to these newer criteria, IVF outcome was suboptimal when normal sperm morphology was os 14% and worst when the proportion became <4%. These observations were substantiated by prospectively designed, blind studies comparing WHO criteria (1987) and the stricter criteria for normal sperm morphology (4, 6), which suggested that the latter method was a superior predictor offertilizing potential (4,6) and had the added advantage of enhanced objectivity in sperm morphological assessment and reduced intertechnician and intratechnician variability (6). However, this view was not shared universally. As with simple morphological assessment, the stricter criteria for normal sperm morphology were reported lacking in accuracy as men with < 14% normal sperm morphology had no significant difference in oocyte fertilization and pregnancy rates when compared with those> 14% (12). Successful oocyte fertilization and pregnancies also have been reported in couples with 0% normal sperm morphology (13). One drawback of attempts to classify sperm into morphological subgroups is that each individual sperm is classified once only but may have several deformities. Priority traditionally has been given to deformities of the sperm head over those of the midpiece and to deformities of the midpiece over those of the tail. Accordingly, the sperm deformity index was devised as a method by which the whole sperma- tozoon is assessed by the stricter criteria for normal sperm morphology and classified more than once if more than one deformity exists. Both normal and abnormal sperms are considered and the average number of deformities per sperm is determined to give a value to the sperm deformity index. In this paper we present the results of a prospective study in which the sperm deformity index was evaluated as an expression of detailed morphological assessment that might predict more reliably the outcome of IVF. The sperm deformity index was compared with previously published indicators ofthe percentage normal sperm morphology assessed by the stricter criteria for normal sperm morphology (10, 11) and the multiple anomalies index, which involves the assessment of only abnormal sperm (14). MATERIALS AND METHODS Couples undergoing IVF treatment in the Reproductive Medicine Unit, Liverpool Women's Hospital between April 1993 and June 1994 were considered for this study. Approval for the project was obtained from the local ethical committee. Female partners at 40 years or older or those with a history of severe endometriosis or polycystic ovary syndrome were excluded. Treatment cycles that resulted in an unfavorable ovarian response (less than two mature 00- cytes collected) also led to exclusion from the study. All nonazoospermic male partners were included irrespective of their semen quality. One hundred fiftyeight couples finally were included in the study and underwent one treatment cycle each. Ovarian Superovulation We used a standard ovarian stimulation protocol incorporating the GnRH analogue buserelin acetate (Superfact; Hoechst, Hounslow, United Kingdom) to down-regulate pituitary-ovarian function. This was administered by nasal insufflation at a dose of 100' JLg, five times a day from day 23 of the cycle. Serum E2 of <30 pg/ml «100 pmolll) and the absence of ovarian follicles> 5 mm confirmed quiescent ovaries. Human menopausal gonadotropin (Pergonal; Serono Laboratories, Welwyn, United Kingdom) then was administered by 1M injection at a dose of 225 IU/d. Follicular growth was monitored by regular ultrasound scan assessment until at least three ovarian follicles had attained a diameter of > 18 mm with appropriate serum E2 levels. Human chorionic gonadotropin (5,000 IU Profasi; Serono Laboratories) was administered by deep 1M injection 35 hours before ultrasound-directed vaginal oocyte retrieval. Oocytes and embryos were cultured according to the standard Reproductive Medicine Unit protocols Vol. 66, No.6, December 1996 Aziz et al. Sperm deformity index 1001

3 r described elsewhere (15). Insemination was performed 40 hours after hcg using 100,000 prepared motile sperm per oocyte. At 56 to 58 hours after hcg 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 but were not considered when accounting for the normally dividing embryos. A maximum of three embryos were transferred 48 hours postoocyte collection. Luteal support, in the form of P vaginal pessaries (Cyclogest; Hoechst), was administered at a dose of 200 mg two times per day for 2 weeks. A pregnancy was defined as the presence of an intrauterine gestational sac containing a fetal pole with the presence of a beating heart as detected by an ultrasound scan. 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 WHO criteria (3). A standard hemocytometer (improved Neubauer) was used to measure concentration and a clean glass microscope slide was used for motility evaluation. Sperm were prepared for insemination by a standard swim-up technique. The sample was divided into two or more Falcon 5-mL tubes (Becton Dickinson, Oxford, United Kingdom) and diluted 1:4 with pre-equilibrated IVF culture medium and centrifuged at 600 X g for 10 minutes. The supernatant was removed and the sperm pellet resuspended in 0.3 to 0.5 ml fresh culture medium, with a further 1 ml layered gently over the suspension. Mter incubation at 37 C in 5% CO 2 for a minimum of 1 hour, the upper 0.7 ml layers were combined and centrifuged at 600 X g for 5 minutes. The sperm pellet was resuspended in 0.2 ml culture medium, assessed for sperm concentration and motility as described above, and the volume required to provide 100,000 motile sperm was calculated. Assessment of Sperm Morphology Thin smears of the well-mixed ejaculated (raw) semen produced for the IVF treatment were prepared in duplicate by placing 2 to 5,uL, depending on the sperm concentration, on clean POlY-L-lysinecoated slides. Thin semen smears facilitated sperm morphology assessment by avoiding sperm cell overlap and ensuring that the sperms were scattered at the same focal depth. Mter drying in air for 3 to 4 minutes the slides were fixed immediately and stained with Spermac fixative and stain (Stain In Aziz et at. Sperm deformity index terprise, Wellington, Republic of South Mrica). This is a modified Papanicolaou stain that stains the sperm acrosomes green, the head equatorial region pale green, and the rest of the head red. The midpiece and tail also stain green (16, 17). Slides were coded and evaluated randomly by the investigator (N.A.) in monthly batches with no knowledge of the coding or the outcome of the treatment. One hundred sperm were studied per slide using bright-field illumination and an oil immersion objective with a total magnification of x2,000. At least 10 high-power fields selected at random from different areas of the slide were examined. A calibrated micrometer on the eyepiece of the light microscope was used to measure sperm dimensions when there was doubt over sperm classification. All slides were assessed using a morphological classification based on a modification of the method of Eliasson (18) and applying the stricter criteria for normal sperm morphology (5,10). Sperm was considered normal when the head had a smooth oval shape with a well-defined acrosome covering 40% to 70% of its apical part. The head length was required to be 3 to 5,urn and width 2 to 3,urn. In addition, the midpiece was required be axially attached, slender, :51,urn in width and approximately 1.5 times the head length. Any cytoplasmic droplet present had not to exceed half the head area. The tail also was required to be uniform, slightly thinner than the midpiece, uncoiled, free from kinks, and approximately 45,urn in length. The morphologically abnormal sperms were classified into subgroups that included pyriform, tapered, largeheaded, small-headed, acrosomal defects, amorphous, cytoplasmic droplet> 0.5 the head size, midpiece defects, tail anomalies, and double sperms (double head or tail in any combination). Borderline forms were considered abnormal and included those sperms with slightly elongated head with loss of its oval shape, those with rounded heads and intact acrosome, or those with normal heads and a thickened midpiece. A multiple entry scoring technique was adopted in which an abnormal sperm was classified more than once if more than one deformity was observed. The sperm deformity index was calculated by dividing the total number of deformities observed by the number of sperm randomly selected and evaluated irrespective of their morphological normality. The multiple anomalies index, defined as the average number of abnormalities per abnormal sperm, also was calculated for each semen sample by continuing the morphological evaluation using the stricter criteria for normal sperm morphology until 100 abnormal sperms had been assessed. Quality control of sperm morphology assessment was carried out throughout this study. Fifteen coded

4 semen smear slides were assessed randomly in three-monthly cycles. Two-way analysis of variance of four repeated assessments by one of us (N.A.) revealed no significant difference in repeated estimation of different sperm morphological forms (for normal forms: F = 1.06, P = 0.37, 95% limits of agreement = -1.6 to 0.9; for sperm deformity index: F = 2.2, P = 0.1,95% limits of agreement = -2.7 to 1.5). Statistical Methods The Arcus Biomedical computer software package (Arcusbiomedical, Medical Computing, United Kingdom) was used to assist statistical analysis. The Shapiro-Wilk W test revealed evidence of non-normality of distributions of many of our data and therefore we used nonparametric statistical methods. The Mann Whitney U test was used to compare independent groups involving patient and sperm characteristics. Receiver operating characteristic (ROC) curves for the sperm deformity index, percentage normal sperm morphology, and multiple anomalies index were constructed by plotting the sensitivity against the false-positive rate (I-specificity) at different cutoff points to compare their value as diagnostic tests for fertilization. Sensitivity, specificity, positive predictive value, negative predictive value, likelihood ratio of a positive test and negative test, and the pretest and post-test odds were calculated for our diagnostic test data. Areas under the ROC curves were calculated using the trapezoidal rule with 500 different cutoff points between the observed minimum and maximum (19). Kendall's rank correlation was used to assess the interdependence of semen parameters and oocyte fertilization. RESULTS One hundred fifty-eight couples underwent one treatment cycle each, of which 116 couples (73.4%) achieved fertilization and 42 couples (26.6%) did not. Oocyte fertilization rates varied between 10% and 100% (median 86%). The average 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). Sperm Count and Motility The WHO criteria (3) for normal sperm count (::=:20 X 10 6 /ml) and progressive motility (::=:50% with forward progression) in the raw semen were met by 85% (134/158) of the male partners. Of these 134 patients with normal count and motility, 23 (17%) did not achieve fertilization. Twenty-four patients had oligospermia and/or asthenospermia and, of these, 5 (21%) patients achieved fertilization. The median sperm count and motility in the raw semen were significantly higher in the fertilizers than the nonfertilizers (Table 1), but the degree of overlap between fertilizers and nonfertilizers for each parameter was considerable. All fertilizers had a normal sperm count and only five of this group displayed asthenospermia. Among the 42 nonfertilizers, 23 (55%) had normal sperm count and motility, 6 were oligospermic, 10 displayed asthenospermia, and 3 had oligoasthenospermia. Normal Sperm Morphology There was highly significant interdependence of the proportion of normal forms as identified by the stricter criteria for normal sperm morphology in the raw semen and oocyte fertilization rates (Kendall's tau b = 0.27; P < 0.001). The fertilizers showed a significantly greater median percentage normal sperm morphology compared with the nonfertilizers (Table 2). The normal sperm morphology scores for the nonfertilizers ranged between 0% and 17%, whereas the fertilizers always displayed some sperms with normal morphology in their raw semen smears (range 1% to 50%). Abnormal Sperm Morphology A comparison of the proportion of different sperm morphological subgroups in the raw semen of the fertilizers and the nonfertilizers is given in Table 2. The fertilizers had a significantly greater median Table 1 Comparison of Patients' Characteristics and Sperm Count and Motility Between the Fertilizers Fertilizers (n = 116) Non-fertilizers (n = 42) 95% CI for difference Two- Criteria Median Range Median Range between medians sided P Female age (y) to to 39-1 to Period of infertility (y) 6 1 to to 16-1 to No. of oocytes inseminated 5 2 to to 13-1 to Sperm count (10 6 /ml) to to to -31 <0.001 Sperm motility (%) to to to -9 <0.001 Vol. 66, No.6, December 1996 Aziz et at. Sperm deformity index 1003

5 .. Table 2 Different Morphological Forms as Defined by the Stricter Criteria for Normal Sperm Morphology in the Raw Semen of Fertilizers Compared With Nonfertilizers Fertilizers (n = 116) Morphological forms Median Range Normal forms (%) 11 lto 50 Borderline (%) 22 3 to 42 Amorphous (%) 31 4 to 80 Pyriform (%) 5 to 44 Taper (%) 7 to 40 Large (%) 2 to 17 Small (%) 3 to 53 Double (%) 0 to 15 Cytoplasmic droplet (%) 1.5 o to 12 Midpiece defect (%) 13 2 to 38 Tail defect (%) 9 o to 35 Acrosomal defect (%) 16 lto 45 Sperm deformity index to 1.9 Nonfertilizers (n = 42) 95% CI for difference Two- Median Range between medians sided P 3 to 17-9 to -5 < o to 36-9 to -3 < to 72 5 to 15 < to 48-1 to to 30-2 to to 12-2 to to 61 to to 8 to o to 19 o to to 59 3 to 9 < to 51 1 to to to 19 < to to 0.34 <0.001 borderline sperms and a significantly smaller median amorphous sperms, midpiece defects, cytoplasmic droplets, tail defects, and acrosomal defects compared with the nonfertilizers. There was considerable overlap between fertilizers and nonfertilizers for each of these parameters. Sperm Deformity Index There was a highly significant negative interdependence of sperm deformity index and fertilization rate (Kendall's tau b = 0.32; P < 0.001). Fertilizers Sensitivity had a highly significantly smaller median sperm deformity index than nonfertilizers (Table 2). Moreover, contrary to other semen parameters studied, there was a minimal overlap of the sperm deformity index between the two groups. This allowed the selection of a cutoff point that would be useful clinically in predicting the outcome of oocyte fertilization. The ROC curve was constructed and a cutoff point at sperm deformity index 1.6 was that which yielded the maximum sum of sensitivity and specificity (Fig. 1). Sperm deformity index :s; 1.6 was considered a positive test compatible with fertilization, and> 1.6 a negative test predicting nonfertilization. At this cutoff point, the sperm deformity index test correctly predicted fertilization in 111 of 116 patients achieving fertilization (sensitivity = 96%). It also identified 30 of 42 patients who did not fertilize (specificity = 71%). The positive predictive value (likelihood of fertilization after a positive test) was 90%, and the negative predictive value (likelihood of failed fertilization after a negative test) was 86%. The likelihood ratio of a positive test was 3.35 (95% Gonfidence interval [CI] 2.2 to 5.6) and the likelihood ratio of a negative test was 0.06 (95% CI 0.03 to 0.1). Having a positive sperm deformity index test (:s;1.6) increased the odds in favor of achieving fertilization to 9 to 1. Conversely, the odds of not achieving fertilization after a negative test (> 1.6) were 6 to 1 (Table 3). Sperm Deformity Index and Percentage Normal Sperm Morphology 1-Specificity Figure 1 The ROC curve for sperm deformity index, normal sperm morphology, and multiple anomalies index demonstrating the cutoff point for each parameter on the curve and the larger area under the sperm deformity index test ROC curve compared with the other two Aziz et ai. Sperm deformity index To examine how well the sperm deformity index test performed as a prognostic test compared with the estimation of the proportion of sperm with normal morphology in the raw semen, we compared their respective ROC curve (Fig. 1). The area under the ROC curve was greater for the sperm deformity

6 Table 3 Comparison of the Sperm Deformity Index, Percentage Normal Sperm Morphology and Multiple Anomalies Index as Predictors of the Outcome of Oocyte Fertilization at the Selected Cutoff Points Test property Sperm deformity index test cutoff point 1.6 Normal sperm morphology test cutoff point 4% Multiple anomalies index test cutoff point 1.75 Sensitivity (%) Specificity (%) Positive predictive value (%) Negative predictive value (%) Likelihood ratio of a positive test* Likelihood ratio of a negative test* Odds in favour of fertilization given a positive test Odds in favour of failed fertilization given a negative test (2.2 to 5.6) 0.06 (0.02 to 0.13) 9.3 to 1 6 to (1.9 to 4.6) 0.2 (0.1 to 0.3) 8 to 1 2 to (1.4 to 2.3) 0.06 (0.02 to 0.2) 5 to 1 6 to 1 * Values in parentheses are 95% CI. index (0.875) than for percentage normal sperm morphology (0.622). A cutoff point for normal sperm morphology was identified at 4%, which maximized the sensitivity and specificity. A normal sperm morphology > 4% was considered a positive test and :5;4% as a negative test. When compared, the sperm deformity index test had a higher sensitivity and specificity than normal sperm morphology (Table 3). As a result, a positive sperm deformity index test gave higher odds in favor offertilization than the percentage normal sperm morphology. In addition, the sperm deformity index test provided a greater certainty than percentage normal sperm morphology in predicting the potential nonfertilizers (negative predictive value 86% and 66%, respectively). Moreover, the sperm deformity index test had a larger likelihood ratio of a positive test and a smaller likelihood ratio of a negative test, reflecting the greater accuracy of the test compared with percentage normal sperm morphology estimation. Sperm Deformity Index and Multiple Anomalies Index The multiple anomalies index had a highly significant negative correlation with oocyte fertilization rate (Kendall's tau b = -0.3; P < 0.001). The value of the multiple anomalies index in predicting the outcome of IVF also was examined by constructing its ROC curve. The area under the ROC curve was greater for the sperm deformity index (0.875) than for the multiple anomalies index (0.852). In contrast to the sperm deformity index and percentage normal sperm morphology, the multiple anomalies index displayed a greater overlap between fertilizers and nonfertilizers. This made it difficult to select a clinically useful cutoff point. The optimized multiple anomalies index cutoff at 1.75 achieved a sensitivity similar to the sperm deformity index test but had poor specificity (Table 3). At this optimized cutoff point, the multiple anomalies index test was a good Vol. 66, No.6, December 1996 negative test with negative predictive value similar to that of the sperm deformity index test and superior to that ofthe percentage normal sperm morphology test. Sperm Morphology and Poor Fertilization Of the 116 fertilizers, a group of 12 patients had poor oocyte fertilization rates of <50% (range 10% to 45%; median 34%). Compared with the nonfertilizers, this group of poor fertilizers had a significantly greater median percentage normal sperm morphology (medians 7.5% and 3%, respectively; median difference 4; CI 1 to 6; P = 0.024) and a significantly smaller median sperm deformity index (median 1.55 and 1.65, respectively; median difference -0.18; CI 0.32 to -0.05; P = 0.008) in the raw semen. Compared with patients with fertilization rate > 50%, the poor fertilization group had a significantly smaller median percentage normal sperm morphology (medians 7.5% and 11%, respectively; median difference -4; CI -8 to -1; P = 0.03) and a significantly greater median sperm deformity index (median 1.55 and 1.4, respectively; median difference 0.1; CI 0.2 to 0.02; P = 0.02). The proportion offertilized oocytes that developed into embryos in the two groups of fertilizers was similar (median 100% for both groups; median difference 0; P = 0.18). Sperm Morphology and Pregnancy Outcome Of 116 couples who achieved fertilization, 111 couples had embryos available for transfer and these 111 transfers resulted in 26 pregnancies. Two of the pregnancies were ectopic and were disregarded in any further statistical consideration. Four pregnancies miscarried between 9 and 13 weeks gestation, and 20 pregnancies resulted in the birth of 14 singletons and 6 sets of twins. The clinical pregnancy rate was 15.2% per cycle and 21.6% per ET. The proportion of the various sperm morphological forms as- Aziz et al. Sperm deformity index 1005

7 sessed and the sperm deformity index were not significantly different in couples achieving pregnancy and those who did not. Moreover, there was no statistically significant interdependence between any sperm morphological form and the clinical outcome of the pregnancy. DISCUSSION This paper describes the sperm deformity index as a novel expression of sperm morphological assessment by the stricter criteria for normal sperm morphology and reports its correlation with fertilization rate and its value as a predictive test superior to previously published indicators such as normal sperm morphology using the stricter criteria for normal sperm morphology and the multiple anomalies index. The value of the sperm deformity index appears to be in the minimal degree of overlap between the group of patients achieving fertilization and those who did not, allowing the selection of a clinically useful cutoff point yielding high sensitivity and specificity and high positive and negative predictive values compared with either the percentage normal sperm morphology test or the multiple anomalies index test. It identified the vast majority of potential nonfertilizers. These patients could be offered alternative techniques, such as intracytoplasmic sperm injection (les!), to maximize their chances of successful treatment. Moreover, the high negative predictive value minimizes the number of patients wrongly predicted as potentially nonfertilizers and unnecessarily offered the more labor intensive IeSI treatment. The sperm deformity index test can be performed quickly by a trained technician using only 3 to 5 ILL of the raw semen produced for IVF. It uses an inexpensive 5-minute staining technique, requires no additional equipment, and is quicker to calculate than the multiple anomalies index, which requires the continuation of the sperm morphological assessment to include the desired number of abnormal sperms. The use of raw semen rather than the sperm preparation allows adequate time to discuss alternative assisted conception techniques, if required. Rationalizing the indications for IeSI thus helps to boost the IVF program success rate within a manageable laboratory workload. The positive correlation between the percentage normal sperm morphology assessed by the stricter criteria and fertilization rates has been reported previously (5-8, 10, 11, 20) and was evident in our study. The ROe curve produced a clear prognostic cutoff point at 4%. The odds in favor of fertilization when the normal sperm morphology is in excess of 1006 Aziz et al. Sperm deformity index 4% are 8 to 1, which renders the normal sperm morphology test a good positive predictor. However, one third of patients with normal sperm morphology score :54% achieved fertilization, making the normal sperm morphology test a relatively poor negative predictor. In contrast, the multiple anomalies index at the optimized cutoff point (> 1.75) was a good negative test for fertilization with negative predictive value similar to that of the sperm deformity index test (85% and 86%, respectively). However, a positive multiple anomalies index test «1. 75) gave odds in favor of fertilization that were inferior to those of the normal sperm morphology or the sperm deformity index. The current lack of a consensus on the value of sperm morphology in predicting the outcome of IVF treatment (1,2,4-8, 10-13) has been attributed to a host of factors. First, sperm morphology assessment methods in different assisted conception units are not standardized (21). Second, the design and study populations for sperm morphology evaluation are variable, making it difficult to compare results. Third, although it has been shown that fertility declines with declining normal sperm morphology, this criterion does not define a specific pathology and therefore its relationship is continuous rather than discrete. Hence, any arbitrary cutoff of normality of sperm morphology simply isolates the extremely low percentile values in an otherwise continuous distribution. A continuous relationship between oocyte fertilization rates and sperm morphology was quite evident in our study group, as demonstrated by the graduated influence of percentage normal sperm morphology and sperm deformity index on oocyte fertilization rate. However, it also was clear that, once fertilization had occurred, cell division, embryo development, and the establishment of pregnancy were independent of these morphological patterns. In this study, patient selection bias allowed for by other workers (5, 6, 10, 11, 20) w~s avoided by including all nonazoospermic males whose female partner did not have any of the conditions known to compromise oocyte fertilization. This allowed us to assess the relative value of other standard sperm parameters (count and motility) in predicting oocyte fertilization outcome in real clinical situations presenting a spectrum of problems encountered in the male. Although patients achieving fertilization had significantly higher sperm counts and progressive motility than those who did not, the predictive value of either parameter was compromised by the considerable overlap of the absolute values seen in both groups of patients. This supports the general consensus that these simple parameters are insufficient for a reliable prediction of the outcome of oocyte fertilization.

8 1 I Historically, male fertility initially was associated with the proportion of abnormal sperm in semen samples (22, 23). It was believed that a high proportion of abnormal forms may be a reflection of pathological spermatogenesis and that coexisting sperm seen as normal therefore may possess subtle unrecognizable defects compromising their fertilizing ability. However, the emphasis has gradually shifted and the proportion of normal forms more recently has been used instead to determine the fertility potential of a semen sample. This strategy was developed to overcome the confusion caused by classifying abnormal sperm forms into as many as 70 different subgroups (24). These subgroups existed in small percentages and, with few exceptions, the clinical value of the tedious process of subgrouping was debatable (18, 24, 25). In addition, the practice of classifying sperm with multiple deformities only once with priority given to deformities of the sperm head led to the underestimation of deformities of sperm mid piece and tail, the integrity of which is indispensable for the spontaneous fertilizing competence of the sperm. The sperm deformity index, by permitting the classification of the same sperm more than once if more than one abnormality exists, ensures that deformities of different parts of the sperm are accounted for equally. This provides scope for the evaluation of the influence of specific sperm structural deformities, singly or in combinations, on sperm-oocyte interaction. The sperm deformity index also may resolve the long- standing difficulty some observers have had regarding the correct classification of sperm heads with moderate postequatorial narrowing (23). Their classification as either a normal variation (18, 23) or as a borderline abnormality (5, 10) would contribute the same value to the sperm deformity index. The absolute value of the sperm deformity index reflects the interplay between the proportion of sperm with normal morphology and the proportion with multiple deformities as seen in a fixed stained semen smear. The index, in turn, incorporates the undeniable predictive value of normal sperm morphology. At sperm deformity index> 1.6, the fertilizing potential of the semen sample is compromised despite the presence of normal forms. This implies that these morphologically normal sperms may be nevertheless functionally abnormal or have an abnormal ultrastructure induced by the same factor(s) that gave rise to the high frequency of sperms with multiple deformities. In conclusion, the sperm deformity index is a novel expression of sperm morphological parameters, the absolute value of which reflects the balance between the prevalence of sperms with multiple structural deformities and the proportion of sperm with normal Vol. 66, No.6, December 1996 morphology in a semen sample. In our study population, we found the sperm deformity index to be a more reliable predictor of the outcome of oocyte fertilization in vitro than either the percentage normal sperm morphology or multiple anomalies index. In this study, a sperm deformity index> 1.6 was found to reduce the fertilizing potential of a semen sample even when normal sperms were observed in the stained smear. We hypothesize that, at this cutoff point, morphologically normal sperm as seen under the microscope may incorporate some as yet unrecognizable defects that compromise their fertilizing ability. As a reliable predictor, the sperm deformity index allows the rationalization of more expensive and labor intensive alternative treatments such as ICSI, assisting to keep laboratory work within a manageable level. The test is quick and inexpensive and can be performed on the ejaculate produced for the IVF treatment, thus giving an up-to-date assessment of sperm morphology. The multiple entry technique avoids the current confusion in classifying sperm displaying more than one deformity. Finally, as the study population is relatively small, it is important to validate our findings by a multicenter trial to confirm the predictive value of sperm deformity index and to assess the intraobserver and interobserver variability. REFERENCES 1. 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: Trounson A, Wood C. In vitro fertilization results, , at Monash university, Queen Victoria, and Epworth Medical centres. J In Vitro Fert Embryo Transf 1984; 1: World Health Organization. Laboratory manual for examination of human semen and sperm-cervical mucus interaction. 2nd ed. Cambridge: The Press Syndicate ofthe University of ; Cambridge, Oehninger S, Acosta AA, Morshedi M, Veek L, Swanson RJ, Simmons KF, et al. Corrective measures and pregnancy outcome in IVF patients with severe sperm morphologic abnormalities. Fertil Steril 1988;50: Menkveld R, Stander FSH, Kotze TJvW, Kruger TF, Van Zyl JA. The evaluation of morphological characteristics of human spermatozoa according to strict criteria. Hum Reprod 1990; 5: Enginsu ME, Dumoulin J, Pieters M, Bras M, Evers J, Geraedts J. Evaluation of human sperm morphology using strict criteria after Diff-Quick staining: correlation of morphology with fertilization in vitro. Hum Reprod 1991;6: Hinting A, Comhaire F, Vermeulen L, Dhort M, Vermeulen A, Vanderberhove D. Value of sperm characteristics and the result of in vitro fertilization for predicting the outcome of assisted reproduction. Int J Androl 1990; 13: Duncan WW, Glew MJ, Wang X, Flaherty SP, Matthews CD. Prediction of fertilization rates from semen variables. Fertil Steril 1993;59: Aziz et ai. Sperm deformity index 1007

9 9. Williams WW, Savage A. Observations on the seminal morphology of bull. Cornell Vet 1925;15: 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: Kruger TF, Acosta AA, Simmons KF, Swanson RJ, Matta JF, Oehninger S. Predictive value of abnormal sperm morphology in in vitro fertilization. Fertil Steril1988;49: Morgentaler A, Fung MY, Harris DH, Powers RD, Alper MM. Sperm morphology and in vitro fertilization outcome: a direct comparison of World Health Organization and strict criteria morphologies. Fertil Steril 1995; 64: Seibel MM, Zilberstein M. The diagnosis of male infertility by semen quality: the shape of sperm morphology. Hum Reprod 1995; 10: Jeulin C, Feneux D, Serres C, Jounnet P, Guillet-Rosso F, Belaisch-Alart J, et al. Sperm factors related to failure of human in vitro fertilization. J Reprod Fertil 1986; 76: Kingsland CR, Aziz N, Taylor CT, Manasse PR, Haddad N, Richmond DH. Transport in vitro fertilization-a novel scheme for community-based treatment. Fertil Steril 1992; 58: Oettle EE. Using a new acrosome stain to evaluate sperm morphology. Vet Med 1986;81: Menkveld R, Oettle EE, Kruger TF, Swanson RJ, Acosta AA, Oehninger S. Atlas of human sperm morphology. Baltimore: Williams & Wilkins, 1991:P Eliasson R. Standards for investigation of human semen. Andrologie 1971;3: Metz CEo Basic principles of ROC Analysis. Semin Nucl Med 1978;8: Grow D, Oehninger S, Seltman HJ, Toner JP, Swanson RJ, Kruger TF, et al. Sperm morphology as diagnosed by strict criteria: probing the impact of teratospermia on fertilization rate and pregnancy outcome in a large in vitro fertilization population. Fertil Steril 1994;62: Barrat CLR. On the accuracy and clinical value of semen laboratory tests. Hum Reprod 1995; 10: Moench GL, Holt H. Sperm morphology in relation to fertility, Am J Obstet Gynecol 1931;22: Hotchkiss RS, Brunner EK. Semen analysis of two hundred fertile men. Am J Med Sci 1938; 196: Macleod J. The semen examination. Clin Obstet Gynecol 1965;8: Makler A. Modern methods in semen analysis evaluation. In: Behrman SJ, Kistner RW, Paton GW, editors. Progress in infertility. Boston: Little Brown and Company, 1988: Aziz et al. Sperm deformity index