Human sperm acrosome function assays are predictive of fertilization rate in vitro: a retrospective cohort study and meta-analysis

Objective To determine whether acrosome function scoring—including acrosomal enzyme (AE) levels and acrosome reaction (AR) results—can predict fertilization rate in vitro. Methods We examined the predictive value of acrosomal enzymes (AE) determined by spectrophotometry/N-α-benzoyl-dl-arginine-p-nitroanilide for fertilization rate (FR) in vitro in a retrospective cohort study of 737 infertile couples undergoing IVF therapy. Additionally, a meta-analysis was done for prospective cohort or case-control studies; the following summary measures were reported to expand upon the findings: pooled spearman correlation coefficient (Rs), standardized mean difference (SMD), sensitivity (SEN), specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic score (DS), diagnostic odds ratio (DOR), and area under the summary receiver operating characteristic curve (AUC). Results Lower AE levels determined by spectrophotometry with a cut-off value of <25μIU/106 spermatozoa were predictive of total fertilization failure (TFF) with moderate SEN (88.23%) and low SPE (16.50%). On meta-analysis, a total of 44 unique articles were selected, but given the multiple techniques described there was a total of 67 total datasets extracted from these 44 articles, comprising 5356 infertile couples undergoing IVF therapy. The AE levels or induced AR% was positively correlated with FR (Rs = 0.38, SMD = 0.79; Rs = 0.40, SMD = 0.86, respectively). Lower AE levels or induced AR% was predictive of lower fertilization rate with moderate accuracy (AUC = 0.78, AUC = 0.84, respectively); this was accompanied by low SEN/moderate SPE (0.57/0.85), moderate SEN/moderate SPE (0.79/0.87), respectively. For AE assay, the diagnostic performance in Asia (Rs = 0.24, SMD = 0.50) was inferior to that in North America (Rs = 0.54, SMD = 0.81) and Europe (Rs = 0.46, SMD = 0.92). Cryopreserved spermatozoa (SMD = 0.20, P = 0.204) were inferior to fresh spermatozoa (SMD = 0.89, P <  0.001). Sperm preparation yielded inferior results as compared to no preparation; spermatozoa after swim up were weak relevant (Rs = 0.27, P = 0.044); and there was no correlation for spermatozoa after a discontinuous gradient (SMD = 1.07, P >  0.05). Lower AE levels determined by fluorometry or substrate assay were used for predicting lower FR with low sensitivity and high specificity; the spectrophotometry assay had an uncertain predictive value. For induced AR assay, the diagnostic performance in the other areas was inferior to that in Africa (Rs = 0.65, SMD = 1.86). No preparation or double preparation yielded inferior results as compared to one preparation (Rs = 0.41); discontinuous gradient (Rs = 0.17, SMD = 0.47) was inferior to swim up (Rs =0.65, SMD = 1.51). Nonphysiological triggers (SMD = 0.81) did not differ from physiological triggers (SMD = 0.95) in general; ZP (Rs = 0.63) or mannose (Rs = 0.59) was superior to other physiological or nonphysiological triggers; and there was no correlation for human follicle fluid, progesterone, cyclic adenosine 3′-5′-phosphate analogue and phorbol ester–BSA-GlcNAc Neoglycoproteins with N-acetylglucosamine residues. Lower induced AR% determined by indirect immunofluorescence, direct immunofluorescence with lection, or triple stain was used for predicting lower FR, with moderate sensitivity/high specificity, moderate sensitivity/high specificity, or high sensitivity/low specificity. Conclusions Although the correlation between acrosome function scoring and FR was significant, the assays were neither highly sensitive nor specific. Additionally, the diagnostic performance showed regional effects as well as an effect of the sperm preparation or assay method. More studies of multicenter, large-scale, careful design and synthesizing multiple sperm functional assays and oocyte quality assays are still needed in clinical settings to better predict fertilization outcome in IVF. Electronic supplementary material The online version of this article (10.1186/s12958-018-0398-y) contains supplementary material, which is available to authorized users.


Background
The sperm acrosome is a Golgi complex-derived flat granule overlaying the anterior two-thirds of the sperm head and contains numerous acrosomal enzymes (AEs) such as protease, glycosidase, acrosin, hyaluronidase, and high-electron density semisolid matrix proteins. Among AEs, the serine proteinase acrosin and hyaluronidase are of particular interest owing to their roles in fertilization, which include limited proteolysis of zona proteins to facilitate spermatozoa penetration into the various layers of the ovum. Acrosin-which is exclusive to the acrosome of mammalian spermatozoa-is mainly synthesized and stored in an enzymatically inactive zymogen form (i.e., proacrosin), and is released during acrosomal exocytosis following maturation [1]. Hyaluronidase is secreted and depolymerizes the matrix between cells of the cumulus oophorus [2].

AE determination
Prior to further inclusion of couples in therapy protocol, the semen samples were collected and AE levels were determined by the procedure of Kennedy [6], with proper modifications. Briefly, the experimental and control tubes, each containing 7.5 × 10 6 spermatozoa, were layered over 500 μL of 11% Ficoll (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged at 2000×g for 20 min. Then 100 μL of benzamidine (500 mM, Sigma-Aldrich, St. Louis, MO, USA) was added to equal volume of sperm pellet in the control tube. Afterwards, 1 mL of substrate-detergent mixture (BAPNA-Triton X-100 mixture, PH = 8.0, Sigma-Aldrich, St. Louis, MO, USA) was added to both tubes. After 1 h of incubation at 24°C, benzamidine (100 μL) was added to experimental tube to stop the reaction. All samples were centrifuged at 2000×g for 15 min and the absorbance of supernatants was spectrophotometrically determined at 410 nm. AE activity (μIU/10 6 ) was calculated out of the difference in optical density between experimental and control tube of each sample.

Meta-analysis Data sources and study selection
Two investigators independently carried out a search in PubMed, Web of Science, Cochrane Library, Embase, EBSCO, Ovid, ClinicalTrials.gov and Google Scholar databases for relevant literature up to February 2017. The [Title/Abstract] search was restricted to English language publications and was performed for the following MeSH terms: fertilization in vitro, acrosin, acrosome reaction, exocytosis, predictive value of tests, sensitivity and specificity (Additional file 1: search strategy). Inclusion criteria were as follows: (1) prospective cohort or case-control design; (2) infertile couples undergoing IVF therapy; (3) a study population of at least 30 couples; (4) AE or AR assay as an index test; (5) oocytes examined to establish fertilization as a reference standard test.

Data extraction and quality assessment
Information on study characteristics was independently abstracted by two investigators according to a standardized table (Table 2-4), with decisions made by consensus in cases of disagreement. In four articles where there were ≥ 1 outcome indicators, data with a maximal correlation coefficient and corresponding 95% confidence interval (CI) were used [16,18,25,44]. In four articles where there were ≥ 1 AE/AR cut-off values, data with the best sensitivity (SEN) or specificity (SPE) were used [3,6,39,59]. The methodological quality of eligible articles was assessed with the QUADAS-2 tool [60]. Based on user guidelines, items were tailored by omitting or modifying some signaling questions [60]; for example, when reviewing Patient Selection, the item "Was a casecontrol design avoided?" was omitted; and for a review of Objective Index Test, the item "If a threshold was used, was it pre-specified?" was substituted with "Was the method of determining AEs or AR described?" This substitution was made because candidate articles were included regardless of the method of acrosome function detection.

Statistical analysis
In retrospective cohort study, the statistical analysis was performed by SPSS version 16.0 for Windows (SPSS Inc., Chicago, IL, USA). Data were presented as number and percentages for categorical variables, while non-normal variables were reported as median and interquartile ranges. Spearman rank analysis was performed to determine which variables were related to FR. The Pearson χ2-test was performed for comparison for the frequencies of categorical variables. Two-tailed p < 0.05 was considered statistically significant. In meta-analysis, data analysis was performed using STATA 12.0 software (Stata Corp., College Station, TX, USA). Statistical heterogeneity was evaluated using the Q test or inconsistency index (I 2 ), with significance set at p < 0.05 or I 2 > 50%, respectively. If heterogeneity existed, the random effects model was adopted; otherwise, a fixed-effects model was selected. SEN and subgroup analyses were carried out to identify suspected sources of heterogeneity. Subgroups were compared with the Q test for heterogeneity [61]. The bivariate mixed effects regression model of midas module in STATA 12.0 was used for calculating SEN, SPE, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic score (DS), diagnostic odds ratio (DOR), and for performing the summary receiver operating characteristic (SROC) curve analysis and drawing Fagan nomogram.

AE assay
Rs was extracted from 10 articles that included a total of 758 infertile couples. A total of 13 datasets were analyzed, including one article each that used three [54] and two [9] sperm preparation methods. AE levels and FRs that were higher and lower than the respective cut-off values were extracted from 12 articles, which included a total of 1037 infertile couples. Of the 16 datasets analyzed, one used two AE assay methods [3] and three used two sperm preparation methods [4,7,9]. Binary accuracy data from 939 infertile couples were extracted       ① spearman correlation coefficient; ② induced AR% for higher and lower FR; ③ binary diagnostic accuracy data as 2 × 2 tables from 10 articles as 2 × 2 tables. We analyzed the 12 datasets, including one paper that used three assay methods [3] ( Table 2). According to a random-effects model, AE levels were positively correlated with FR (Rs = 0.39, 95% CI: 0.18-0.60, p < 0.001), albeit with notable heterogeneity (I 2 = 95.7%, p < 0.001; Table 3; Fig. 2a Fig. 2c-e, Table 7). The Fagan nomogram showed that lower AE levels could be used to predict lower FR when the pre-test probability was 27% (i.e., occurrence rate of patients for whom < 70% fertilization was achieved by IVF in our hospital), with a post-test probability of 59%.

Spontaneous AR assay
Rs was extracted from 3 articles that included a total of 181 infertile couples. The spontaneous AR% and FRs that were higher and lower than the respective cut-off values were extracted from 9 articles, which included a total of 602 infertile couples. Of the 10 datasets analyzed, one used two AR assay methods [37]. Binary accuracy data were extracted from only 3 articles as 2 × 2 tables; the diagnostic summary measures were not pooled, based on the computation of bivariate mixed effects regression model for the lowest threshold of 4 studies (Table 3).

Induced AR assay
Rs was extracted from 12 articles that included a total of 917 infertile couples. A total of 17 datasets were analyzed, including one article each that used five [41] and two [39] AR triggers. Induced AR% and FRs that were higher and lower than the respective cut-off values were extracted from 15 articles, which included a total of 1033 infertile couples. Of the 22 datasets analyzed, one used two AR assay methods [37], another reported five AR triggers [41], and two also mentioned different triggers [39,42]. Binary accuracy data from 953 infertile couples were extracted from 12 articles as 2 × 2 tables. We analyzed the 13 datasets, including one paper that used two triggers [39] (Table 4).
In the subgroup analysis, datasets were stratified according to geographic area, sperm preparation method, AR trigger, and AR assay method (Tables 8, 9 Fig. 4 a Graph of SROC curve for induced AR assay. Lower induced AR% was predictive of lower FR with moderate accuracy (AUC = 0.84). c, e Sensitivity analysis plots for spontaneous (c) and induced AR assay (e). The goodness-of-fit and bivariate normality analyses showed that the bivariate model was moderately robust. The influence analysis and outlier detection identified one outlier for spontaneous AR assay, but there was no outlier was identified for induced AR assay. b, d, f Funnel plots for AE assay (b), spontaneous (d) and induced assay (f). The studies were distributed symmetrically without obvious publication bias in three funnel plots       Table 9 Summary of SMD values for induced AR assay     The included studies were distributed symmetrically without obvious publication bias (Deeks' funnel plot [p] = 0.36, Fig. 4f ).

Discussion
There are many functional assays that attempt to assess the fertilization capacity of spermatozoa based on hypoosmotic swelling, peroxidative damage, acrosome status, AEs, sperm chromatin, sperm-oocyte interaction, zona pellucida binding, and zona-free oocyte penetration [65]. However, their clinical utility for diagnosing male infertility is unclear. One reliable criterion for evaluating the diagnostic performance of assays is whether or not they can predict fertilization outcomes in IVF [3,66,67].
Our first results showed that AE (i.e., proacrosin and acrosin) levels determined by spectrophotometry/ BAPNA were positively correlated with FR. However, lower AE levels were predictive of TFF with moderate SEN but with low SPE. In addition, a meta-analysis of published literature on similar topic was performed to further expand upon the findings. To the best of our knowledge, this meta-analysis is the first study to evaluate the association between acrosome function scoringincluding AE levels and AR%-and FR and the diagnostic performance of acrosome function scoring. No attempt has been made here to correlate the scoring with conception rates because several other factors, such as the endometrial secretions, receptivity and systemic and local endocrine status, become significant after embryo transfer [54]. After validating the correlation with pooling Rs and SMD, lower AE levels or induced AR% was predictive of lower FR with moderate accuracy (AUC between 0.70-0.90); this was accompanied by low SEN/ moderate SPE, moderate SEN/moderate SPE, respectively. A moderate SPE indicates that a male diagnosed as scoring -negative (i.e., higher than the AE cut-off value) has about 85% or greater probability of having a high FR (i.e., higher than the FR cut-off value). Fifteen percent of the patients with high AE levels and poor fertilization probably have defects other than impaired AEs [18]. For induced AR assay, the findings were in agreement with the results of Oehninger et al. [68], who reported that AR results were predictive of IVF rates, showing moderate accuracy, SEN and SPE. However, for AE levels, a low SEN indicates that a male diagnosed as AE-positive (i.e., low than the AE cut-off value) still has a 43% probability of having a high FR. The described first results are expected as proacrosin/acrosin is an important enzyme for fertilization. However, the SPE is low, probably because its action is dependent on structural and biochemical events which take place during capacitation and the acrosome reaction and it cannot be detected in its proper location (i.e., the acrosome) like fluorometry [3]. The other kinds of AEs, such as hyaluronidase, were not taken into consideration. Furthermore, the satisfying diagnostic performance was not obtained for assays in the meta-analysis, in spite of synthesizing multiple assay methods. This result in relatively low SEN might be attributed to other parameters of sperm function, such as good membrane integrity, normal chromatin decondensation, excellent ability of undergoing capacitation and hyperactivation, high inducibility of the acrosome reaction (AR), increased sperm-oolemma interaction, or mild peroxidative damage, low DNA fragmentation. However, it should be mentioned that the fertilization process is a multifactorial process where female factors, such as young woman, maturity of oocyte/spindle/zona pellucida, intactness of cumulus-oocyte complex, or good ability to modulate/restore sperm functions, may contribute to high fertilization [15,16]. For spontaneous AR assay, a weak correlation was obtained when pooling Rs; however, after enlarging the sample size, there was no significant correlation between them when pooling SMD. The spontaneous AR assay was considered for the evaluation of the initial acrosome stability before ZP binding; a low percentage of spontaneous AR did not seem to influence sperm fertility may due to high heterogeneity of spermatozoa.
In addition, there was notable heterogeneity when pooling summary measures in the present meta-analysis. After SEN analysis, two studies were identified as a source of heterogeneity when pooling Rs for AE assay. One reported a linear correlation between AE and the percentage of cases with ≥70% fertilization achieved by IVF [8]. On the other hand, semen prepared by α-chymoytrypsin treatment was suitable for highly viscous semen [54]. When SMD was pooled, four studies [3,6,50,53] were found to contribute to this heterogeneity. Two used cryopreserved spermatozoa to assay AE [50,53]; one used spermatozoa without preparation [6] or spermatozoa subjected to a special discontinuous gradient (i.e., 1-ml fractions of 90%, 80%, and 50% Percoll in isotonic Ham's-F10) [3] in IVF therapy. When pooling diagnostic accuracy data, one outlier may have affected inter-study heterogeneity, for which the highest AE cut-off value was obtained by the spectrophotometry/ BAPNA assay (54 μIU/10 6 spermatozoa) [57]. The sperm origin (fresh or cryopreserved), sperm preparation methods, FR cut-off values, and AE assay methods and cut-off values might contribute to inter-study heterogeneity. For spontaneous AR assay, three studies found to contribute to this heterogeneity when pooling SMD. Two used FITC-PSA to determine AR after incubation for 60 min in synthetic human tubal fluid (HTF) media [33,63]; one used two-color fluorescence staining of FITC-PSA and anti-CD46 antibody (MH61) to assay acrosomal status after 4 h of incubation in mBWW/3.5% HSA media [29]. The sperm capacitation time, media, and assay methods might contribute to inter-study heterogeneity. For induced AR assay, seven studies [31,33,41,43,63] were identified as a source of heterogeneity. The inconsistencies among studies regarding capacitation time (range between 1 h and 24 h), sperm preparation methods (swim up or discontinuous gradient), AR triggers (physiological [HFF, P, ZP] or nonphysiological [TPA, CAMP, mannose]), as well as AR assay methods (FITC-PSA, RITC-PSA, FITC-GB24) methods might contribute to inter-study heterogeneity.
Furthermore, the subgroup analysis revealed that the correlation between AE levels and FR depended on geographic area, with Asia being inferior in this regard to North America and Europe, which may be explained by methodological quality. For example, two of three studies in Asia [9,18] did not describe the inclusion criteria for patients undergoing IVF therapy, whereas only a minority of North American (i.e., three of seven) [6,8,19] and European (i.e., one in six articles) [3] studies did not report these criteria. In addition, two Asian studies [9,57] did not clearly define the reference standard test (i.e., fertilization), which was only true for two North American [6,50] and one European [54] study. Additionally, there may be racial differences that could possibly contribute, but this is unknown. The populations of certain areas of the world, such as in parts of North America, can be very heterogeneous as well and racial status cannot be assumed. In the sperm head, the organelle most affected by cryopreservation damage was the acrosome [69], suggesting that cryopreserved spermatozoa were inferior to fresh spermatozoa. Spermatozoa without preparation more closely reflected the population composition and fertility of the original ejaculate [6] and were superior to spermatozoa after swim up and a discontinuous gradient in terms of diagnostic performance. It was difficult to predict FR based on AE levels with high accuracy as well as SEN and SPE using any one assay method. The lower AE levels determined by fluorometry-including pAb-acrosin, pAb-hyaluronidase, and mAb 4D4-proacrosin-could predict TFF with low SEN and high SPE. Lower AE levels determined by the gelatine substrate assay could predict lower FR (i.e., FR ≤ 50% or < 50%) with low SEN and high SPE. As for the hyaluronidase target with agar/hyaluronic acid mixture substrate assay, the diagnostic performance was not evaluated because the described high SEN (0.91) and SPE (1.00) for predicting TFF in the text has contradiction with the calculated low SEN (0.54) and high SPE (1.00) from scatterplot of correlation between hyaluronidase activity and FR in the study by Abdul-Aziz et al. [19]. More studies are needed to determine its predictability. The spectrophotometry assay had an uncertain predictive value. Specifically, the lower AE levels determined by the most commonly used spectrophotometry/BAPNA assay could predict a FR < 70%, FR < 50%, or FR = 0%; this was accompanied by moderate SEN/moderate SPE, pooled low SEN/moderate SPE and moderate SEN/low SPE, respectively. This result also validated the finding from retrospective study. AE levels determined by Accu-Sperm spectrophotometry/BAPNA could predict TFF with low SEN and moderate SPE. However, the lower AE levels obtained by spectrophotometry/BAEE in one study were not correlated with TFF. Another study [10] that was not included in our analysis showed similar results by the same method (AE extraction with acid [i.e., pH = 2.8]) but did not reflect the actual levels of proacrosin converted to acrosin.
For induced AR assay, the diagnostic performance also showed regional effects; the Africa in this regard was superior to other areas, which may be explained by methodology or high inter-study heterogeneity in other certain areas. For example, all three studies [31,35,63] in Africa used the same sperm preparation method (swim up), trigger (ZP), and assay method (FITC-PSA) and clearly defined the reference standard test. Two of them executed the laboratory quality control for assay method by establishing intra-and interassay/technician coefficients of variations, but only one study in other area did [34]. The spermatozoa after one preparationespecially swim up-show better survival after incubation in capacitation media compared with no-prepared or double-prepared spermatozoa, which may explain its optimal diagnostic performance [70]. The nonphysiological triggers did not differ from physiological triggers in terms of diagnostic performance; the mannose maybe act as a substitute when lack of physiological triggers. Nevertheless, the use of human ZP, biologically active recombinant ZP3 or active, synthetic ZP3 peptides (or analogues) combined with a better understanding of the biochemistry of the carbohydrate-protein interactions that take place during gamete recognition, binding and induction of acrosomal exocytosis will undoubtedly help in their elaboration [68]. Finally, it was difficult to predict FR based on induced AR% with high accuracy as well as SEN and SPE using any one assay method. Multiple methods (i.e., indirect immunofluorescence, direct immunofluorescence with lection, and triple stain) may be combined to obtain high SEN and SPE.
In conventional IVF therapy, one of the major disappointments that infertile couples may encounter is the unexpected failure to achieve fertilization. Some researches using early rescue ICSI procedure performed 4-6 h post-insemination have described successful salvage of some total or near-total fertilization failure cycles [71,72]. Therefore, it may provide more important clinic direction when the acrosome function assays were used for predicting TFF. For AE assay, lower AE levels determined by spectrophotometry/BAPNA, Accu-Sperm spectrophotometry/BAPNA, or fluorometry-including pAb-acrosin assay, pAb-hyaluronidase, and mAb 4D4-proacrosin-were used for predicting TFF, with moderate SEN/low SPE, low SEN/moderate SPE, or low SEN/high SPE. For induced AR assay, lower induced AR% determined by triple stain or direct immunofluorescence with lection-including FITC-PSA and FITC-PNA-was used for predicting TFF, with high SEN/low SPE and moderate SEN/moderate SPE. Based on optimal diagnostic performance, a two-method assay using AE levels determined by pAb-acrosin assay and induced AR% determined by triple stain can be recommended for assessing acrosome function and predicting TFF. Two-method assay will reveal four types of detection results: AE levels-postive (< 60% for normal fluorescence scores)/induced AR%-positive (< 31.3% for difference between induced AR minus the spontaneous AR results), AE levels-negative (≥ 60% for normal fluorescence scores)/induced AR%-negative (≥ 31.3% for difference between induced AR minus the spontaneous AR results), AE levels-positive/induced AR%-negative, and AE levels-negative/induced AR%-positive. The early rescue ICSI procedure should be recommended for the patients diagnosed as AE levels-postive/induced AR%-positive, for which has a higher chance of TFF, or patients with high-risk factors-such as unexplained infertility or primary infertility with longer infertility duration-and with conflicting diagnosis (i.e., AE levels-postive/induced AR%-negative or AE levels-negative/induced AR%-positive). The conventional IVF therapy should be recommended for the patients diagnosed as AE levels-negative/induced AR%-negative, for which has a higher chance of fertilization success, or patients with conflicting diagnosis but without high-risk factors.
Our cohort study has several limitations: First, our dataset was collected retrospectively from a single center in a single geographic area and AE was determined by a single spectrophotometric method. Second, the sample size was not large and only FR was the primary fertilization outcome. The meta-analysis results should be considered in the context of their strengths and limitations. The advantages were as follows: the pooling of multiple summary measures; SEN and subgroup analyses to identify sources of heterogeneity; and low publication bias, which confirmed the reliability of the results. Nonetheless, there were some limitations such as no available RCT; the inclusion of old articles (published between 1988 and 2014) and studies with high heterogeneity; and the omission of some AE assay methods, including acrosin/proacrosin/acrosin inhibitor [12] or hyaluronidase [13] target with BAEE substrate assay, hyaluronidase target with cytochemical [14] or hyaluronic acid substrate [2] assay; and acrosin target with western blotting [5] or RIA [20], for which articles were lacking.

Conclusions
The results of our study demonstrate that the acrosome function assays used to predict FR with high SEN and SPE are deficient. A limited prediction was obtained for AE assays, even though multiple methods (i.e., fluorometry, spectrophotometry, substrate assays) may be combined. But for induced AR assay, multiple methods (i.e., indirect immunofluorescence, direct immunofluorescence with lection, and triple stain) may be combined to obtain high SEN and SPE. The diagnostic performance showed regional effects as well as an effect of the sperm preparation or assay method. New assays of acrosome function-such as ones utilizing a panel of monoclonal or polyclonal antibodies against acrosome-related proteins-should be developed as a supplement for a more accurate diagnosis of structural and functional defects in the sperm acrosome. In addition, although most fertility centers rather prefer ICSI than IVF as method of treatment for male-factor infertility couples, yet the pace of this decision-making process should slow down, considering the controversy in the potential safety about ICSI. More studies of multicenter, large-scale, careful design and synthesizing multiple sperm functional assays and oocyte quality assays are still needed in clinical settings to better predict fertilization outcome in IVF. The early rescue ICSI procedure should be recommended for the patients with a higher chance of fertilization failure, and the conventional IVF therapy should be recommended for the patients with a higher chance of fertilization success.