Occurrence and developmental consequences of vacuoles throughout preimplantation development

Transcription

1 Occurrence and developmental consequences of vacuoles throughout preimplantation development Thomas Ebner, Ph.D., Marianne Moser, Ph.D., Michael Sommergruber, M.D., Ute Gaiswinkler, M.D., Omar Shebl, M.D., Klaus Jesacher, M.D., and Gernot Tews, M.D. IVF Unit, Women s General Hospital, Linz, Austria Objective: Since little is known about the actual incidence and fate of vacuoles at different stages of development this preliminary study was set up to accurately measure vacuoles and track them to day 5. Design: Prospective study. Setting: Women s General Hospital in Austria. Patient(s): A total of 223 consecutive IVF and intracytoplasmic sperm injection (ICSI) cycles (206 patients). Intervention(s): Accurate measurement of vacuoles. Affected gametes and embryos were cultured individually and tracked until day 5. Main Outcome Measure(s): Size and number of vacuoles, fertilization rate, blastocyst formation rate, blastocyst quality. Result(s): There was a significant relationship between size of the vacuole (cut-off value 14 m) and fertilization (P.05). At zygote stage the incidence of vacuoles was higher (P.01) in ICSI (11.6%) than in IVF (5.3%). Only 32.2% of affected ICSI-embryos reached blastocyst stage on day 5 compared with 53.0% of the normal ones (P.001). In terms of blastocyst formation vacuolization on day 4 (P.001) turned out to be the most severe one. At blastocyst stage inner cell mass was affected less frequently than the trophectoderm (P.05). Conclusion(s): Three types of vacuoles could be identified: (1) those already present at oocyte collection, which develop during maturation (day 0); (2) those artificially created by ICSI (day 1); and (3) those accompanied with developmental arrest (day 4). The later that vacuoles arose, the more detrimental their effect on blastocyst formation. (Fertil Steril 2005;83: by American Society for Reproductive Medicine.) Key Words: Blastocyst formation, blastocyst quality, cytoplasm, oocyte morphology, vacuolization In controlled ovarian hyperstimulation, it is well recognized that not all of the recruited follicles bear oocytes of the same maturation grade and morphological appearance. A diminished follicular vascularization with subsequent hypoxia is most likely responsible for this phenomenon (1), which will result in different developmental competences of the gametes according to their follicular milieu. Affected oocytes may be characterized by a desynchronization of nuclear and cytoplasmic maturation, e.g., maturation of cytoplasm may still be impaired though resumption of meiosis can be achieved (2). At light-microscopic level, the ooplasm of a presumably normal and mature oocyte has a uniform texture, and its organization is reflected by an apparently random distribution of cell organelles (e.g., mitochondria and endoplasmic reticulum) and other cytoplasmic components (e.g., lipid droplets), with the entire subplasmalemmal ooplasm being populated by cortical granules (3). A great variety of cytoplasmic dysmorphism can be observed in in vitro culture. In contrast to a disastrous change in ooplasmic texture (4, 5) the most frequent cytoplasmic abnormalities, such as dark Received September 20, 2004; revised and accepted February 17, Reprint requests: Thomas Ebner, Assoc. Prof. Dr., IVF Unit, Women s General Hospital, Lederergasse 47, A-4020 Linz, Austria (FAX: ; thomas.ebner@gespag.at). inclusions or refractile bodies, do not impair fertilization rate or embryo quality, at least in intracytoplasmic sperm injection (ICSI) (6 8). Nevertheless, ICSI in dysmorphic gametes may result in embryos with reduced developmental capacity (9). In detail, clinical pregnancy rate was found to be severely impaired (10) and early pregnancy loss seemed to be more frequent (11) in cycles with exclusive transfer of embryos derived from dysmorphic oocytes. Because all cytoplasmic abnormalities were pooled in these studies (10, 11) the actual negative influence of certain cytoplasmic features on implantation behavior remains unclear. However, it may be assumed that the more detrimental cytoplasmic features, such as clusters of smooth endoplasmic reticulum (12) or vacuoles (3), are responsible for this dilemma. To date, the latter morphotype has not been studied in detail, though vacuolization is considered to be the only ooplasmic feature that is associated with fertilization failure in both IVF (3) and ICSI (6). Therefore, this prospective study was set up to analyze the actual influence of vacuolization (time of first appearance, number, and size of vacuoles) on preimplantation development until blastocyst stage on day 5. No data concerning implantation behaviour of vacuolized embryos were collected because our transfer policy did not foresee selection of such embryos for transfer /05/$30.00 Fertility and Sterility Vol. 83, No. 6, June 2005 doi: /j.fertnstert Copyright 2005 American Society for Reproductive Medicine, Published by Elsevier Inc. 1635

2 MATERIALS AND METHODS A total of 223 consecutive cycles (206 patients) frequenting our clinic within a period of 4 months in were included in this prospective trial. The mean age of the women was 32.7 years ( 4.6). Almost three-quarters (165 patients) were part of our ICSI program, whereas the remaining 58 patients had a conventional IVF treatment. Semen analysis showed a sperm count of 20 mio/ml in all ICSI patients but 6 (failed IVF), including 13 cases (7.9%) of azoospermia and 19 cases (11.5%) of severe oligoasthenoteratozoospermia ( 1 mio/ml). In the vast majority of patients motility and/or morphology of the spermatozoa were affected as well. In this group with male factor infertility more than one-third (n 60) of the female partners presented with an additional fertility problem. Logically, this share was found to be increased in the IVF group in which more than 90% presented with tubal blockage (n 36), endometriosis (n 12), or polycystic ovaries (n 6). However, the corresponding semen samples were all inconspicuous. Irrespective of the choice of further treatment, controlled ovarian hyperstimulation was performed using either a GnRH-antagonist protocol (n 138) or a long protocol (n 85). In the antagonist protocol, stimulation was started at the beginning of the cycle (day 2) using FSH (Puregon; Organon, Vienna, Austria). After 5 6 days of stimulation, a GnRH antagonist (Orgalutran; Organon) was administered in addition, depending on the presence of a mm follicle. In the long, downregulation was achieved with a GnRH agonist (Suprecur; Aventis Pharma, Vienna, Austria). After an ultrasound scan as well as a blood analysis, stimulation was initiated with human menopausal gonadotropin (Menogon; Ferring, Kiel, Germany). In all patients ovulation was induced with 10,000 IU hcg (Pregnyl; Organon). Oocyte retrieval was carried out transvaginally under ultrasound guidance 36 hours after hcg-driven ovulation induction. All oocyte-cumulus complexes were incubated for 2 4 hours (BM1 medium; NMS Bio-Medical, Praroman, Switzerland) before insemination, the method of which depended on the semen analysis. In ICSI, brief exposure of oocytes (30 seconds) to 80 IU/mL hyaluronidase (MediCult, Copenhagen, Denmark) facilitated mechanical removal of the cumulus cells. This procedure enables an embryologist to accurately evaluate oocyte maturity and morphology and further helps to carefully inject the male gamete (separated by a swim-up technique) into the female gamete, which was done using our routine ICSI technique (8). Special care was taken not to penetrate vacuoles. However, in those oocytes planned to be inseminated with conventional IVF, morphologic screening was not possible on the day of oocyte collection owing to abundantly present cumulus cells. Conventional insemination was done 1 2 hours earlier than in ICSI. Therefore, sperm were used that were separated from the ejaculate using a Zech glass capillary dish (Astro Med Tec, Salzburg, Austria), without exposure to centrifugation stress. At least one time (1 2 hours after insemination) the Zech-chamber was controlled for signs of sperm rejection which would have necessitated a rescue ICSI. In the morning of day 1 (16 18 hours after insemination) regular fertilization, characterized by two pronuclei and two polar bodies, was checked. The following days (days 2 and 3), embryos were checked for the number and size of blastomeres as well as for the presence of fragments, signs of compaction (day 4), and blastocyst development (day 5). Our media of choice (changed daily) were BM1 medium (NMS Bio-Medical) on day 0, Blastassist System Medium 1 (MediCult) until day 3, and Blastassist System Medium 2 in case of extended culture to day 5. All oocytes or embryos were cultured in groups (at least 10 L per conceptus) under oil (MediCult) unless they showed signs of vacuolization. As planned, all oocytes or embryos showing signs of vacuolization were cultured individually from the very beginning (day 0 in ICSI oocytes, day 1 in IVF zygotes). Thus, number and size of the vacuoles could be documented adequately using an imaging and archival software (Octax Eyeware; MTG, Altdorf, Germany). Whenever (day 0 to day 4) a vacuole arose it was measured accurately and it was attempted to follow the further fate of this cytoplasmic inclusion. This proved particularly difficult on days 3 and 5 owing to either massive fragmentation or high cell number. In such cases embryos or blastocysts were carefully rotated by means of the holding pipette to ensure accurate screening for vacuoles. Spontaneous vacuolization was defined as the rate of vacuolization in previously unaffected gametes or embryos. According to the number and quality of embryos on day 2, transfer was scheduled on either day 3 (n 142) or day 5 (n 62) using an Edwards-Wallace catheter (Smiths Industries, Lancing, UK). Because vacuolized embryos or embryos deriving from vacuolized MII oocytes or zygotes were not transferred routinely, institutional review board approval was not sought. A total of 19 cycles (8.5%) had to be cancelled owing to no or low response (n 16), failed IVF (n 2), or assumed contamination (n 1). Statistical comparison was performed using 2 and t-test. Statistical significance was defined as P.05. Receiveroperating characteristic (ROC) curves were calculated to evaluate the prognostic value of vacuole diameter on fertilization. A deviation of 0.5 m. Spontaneous vacuolization is defined as total number of affected embryos minus number of embryos with preexisting vacuoles. RESULTS Significantly more (P.05) ICSI cycles (67/165) were affected by vacuolization compared with IVF cycles (14/58) Ebner et al. Vacuolization throughout development Vol. 83, No. 6, June 2005

3 TABLE 1 Fertilization and cleavage outcome of all oocytes with and without vacuolization (irrespective of number and size). IVF ICSI n cycles MII oocytes 1,198 Total vacuolization in 47 (3.9) gametes 2Pn (total) 264 (61.7) 774 (64.6) Total vacuolization in 14 (5.3) a 90 (11.6) a zygotes Number of cleavages Total vacuolization d2 21 (8.2) 92 (12.1) Total vacuolization d3 26 (10.2) 104 (13.7) Note: For conventional IVF, oocyte morphology could not be evaluated. a P.01. Ebner. Vascuolization throughout development. Fertil Steril In total, 36.6% of all patients had at least one embryo showing at least one vacuole during preimplantation development. Out of 17 patients who had 2 treatment cycles within the study period only 3 showed recurrent vacuoles (17.7%). As indicated in Table 1, fertilization rate did not differ between IVF and ICSI. Occurrence of vacuoles was not found to be related to patient parameters, controlled ovarian hyperstimulation details, male factor, and hormonal situation at the time of ovulation induction (Table 2). Day 0 (ICSI Only) Though ovulation had been induced, 140/1,358 oocytes (10.3%) showed a germinal vesicle, and 20 (1.5%) were in metaphase I at the time of evaluation. In prophase I oocytes, only one vacuole (0.7%) could be documented. Thus, 1,198 MII oocytes were injectable. Out of 47 mature ICSI oocytes (3.9%) showing vacuolization (Fig. 1A), 31 had single (66%) and 16 had two (21.3%) or more (12.7%) vacuoles. The corresponding fertilization rates (2Pn) were 51.6% for solitary and 43.8% for multiple vacuoles. However, the presence of one or more vacuoles at oocyte stage led to a significantly reduced number of 2Pn zygotes (48.9%) in ICSI compared with unaffected gametes (65.3%) (P.05). Taking into consideration only oocytes with a single vacuole, a significant relationship between the size of the inclusion and the formation of two pronuclei could be observed. In detail, the mean diameter of vacuoles in oocytes which fertilized adequately ( m) was significantly smaller than the diameter ( m) in case of fertilitation failure (P.004). A cut-off value of 14 m for vacuole diameter could be noted above which fertilization did not occur in this study (P.05). Day 1 At zygote stage an overall vacuolization rate of 10% was observed (Fig. 1B), with IVF zygotes showing significantly fewer vacuoles than ICSI zygotes (P.003). None of the vacuoles observed in IVF zygotes disappeared, whereas 10/35 (28.6%) of the spontaneous solitary vacuoles in ICSI oocytes were not detectable on day 2. Another 14 single vacuoles (40%) tended to split. The other vacuoles remained constant in size. Multiple vacuoles at zygote stage never faded away, but more than 80% showed signs of fusion or division on day 2. Days 2 4 From cell stage 2 to 8 (days 2 and 3), spontaneous vacuolization was low in both IVF and ICSI embryos. At compac- TABLE 2 Comparison of patient variables in vacuole-positive cycles and cycles without any embryo showing vacuolization. Vacuole positive Vacuole negative P value Age (y) Number of previous cycles Basal FSH (mu/ml) Stimulation dosage (IU) 1, , E2 (pg/ml) 1,781 1,000 1,465 1, Progesterone (ng/ml) LH (mu/ml) Sperm count ICSI (mio/ml) Sperm count IVF (mio/ml) Note: Hormonal parameters have been controlled on day of ovulation induction (except FSH). Ebner. Vascuolization throughout development. Fertil Steril Fertility and Sterility 1637

4 FIGURE 1 Vacuolization throughout preimplantation development. (A) Vacuole (17 m) in MII oocyte. (B) Ten- m vacuole at zygote stage. (C) Compacting embryo with severe vacuolization (22 35 m). (D) Vacuole (17 m) in the inner cell mass of an expanded blastocyst. CS cytoplasmic string; V vacuole. Ebner. Vascuolization throughout development. Fertil Steril tion stage (day 4), a significant increase in spontaneous vacuolization (Fig. 1C) was observed in IVF (P.005) as well as ICSI (P.001) compared with day 3. In detail, 8.5% of IVF embryos and 10.9% of ICSI embryos showed spontaneous vacuolization on day 4 compared with approximately 2% on day 3. Day 5 Out of 499 embryos (139 IVF embryos and 360 ICSI ones) considered for blastocyst culture, 237 (47.5%) survived until day 5. There was no difference in blastocyst formation rate between IVF (46.8%) and ICSI cycles (47.8%). It has to be mentioned that in IVF the sample numbers were rather small and any influence of vacuolization on blastulation could not be analyzed. However, in ICSI cycles the embryos that showed vacuolization during development had significantly fewer blastocysts (29/90) compared with unaffected embryos (143/270) (P.0006). Further analysis to filter out the most critical day revealed that vacuoles first detected on day 4 (P.0001) have the most severe impact on development to blastocyst stage. Exact localization of vacuoles in blastocysts (n 36) that derived from either IVF or ICSI embryos previously vacuolized was not possible in 11 cases (nondetectable or located in excluded blastomeres). In more than half of the remaining blastocysts (n 17) the vacuoles were detected exclusively in the trophectoderm; twice this happened to the inner cell mass (Fig. 1D). Both cell lines were affected in six day-5 embryos. The inner cell mass (ICM) was affected less frequently than the trophectoderm (P.011) Ebner et al. Vacuolization throughout development Vol. 83, No. 6, June 2005

5 DISCUSSION Vacuolization is probably the most apparent and dynamic cytoplasmic dysmorphism in human oocytes. Vacuoles vary in size as well as in number and, according to Van Blerkom (3), they are membrane-bound cytoplasmic inclusions filled with fluid that is virtually identical with perivitelline fluid. It is assumed that vacuoles arise either spontaneously (3) or by fusion of preexisting vesicles derived from the smooth endoplasmic reticulum and/or Golgi apparatus (13). Vacuoles appear to develop rapidly (within several minutes) around extrusions of the first polar body. Our data support the particular formation between MI and MII stage because only one vacuole-positive prophase I oocyte (germinal vesicle) could be observed during the whole study period. In previous studies, the incidence of vacuolization in mature gametes varied between 5.7% (6) and 12.4% (11). Multiple vacuoles accounted for only a small part of approximately 1% (6, 14). Interestingly, there is only one ICSI study (6) that found a severely impaired fertilization rate in vacuolized oocytes (40%) compared with vacuole-free MII gametes (69.6%). All the other papers either published a normal fertilization rate (11) or did not deal with vacuolization as a separate feature, e.g., combining all kinds of cytoplasmic dysmorphism (8, 10). One possible reason for this divergence might be that all investigators did not take into account the size of the vacuoles. Indeed, the present data are the first to note a specific relationship between size of a vacuole and fertilization outcome of an oocyte (no fertilization in presence of vacuoles 14 m). This phenomenon may be explained by two hypotheses. First, it is very likely that a larger vacuole or multiple vacuoles will cause a much more detrimental effect to the oocyte than a small vacuole because a larger portion of the cytoskeleton (e.g., microtubuli) can not function as it is supposed to. Second, as demonstrated by Van Blerkom (3), large vacuoles can displace the MII spindle from its polar position, which can result in fertilization failure. Even if fertilization occurs in presence of large vacuoles, zygotes may fail to initiate first cleavage or display a rather abnormal cytokinesis pattern (3, 15). It can be estimated, that in the present study the vacuolization rate in IVF oocytes was similar to that in ICSI oocytes after denudation because patient parameters and stimulation details were the same. In conventional IVF it has been reported (16) that endocytotic vacuoles can appear de novo in relatively rare instances, indicating that in the present study most of the vacuoles detected at zygote stage were indeed relics from day 0. This is of particular interest because IVF zygotes were significantly less affected by vacuoles than ICSI zygotes were (P.01). This strongly suggests a direct influence of the ICSI procedure per se in this study, e.g., an encapsulation of polyvinylpyrrolidone or medium which might have gained access to the cell during injection. This would be in line with the finding that ICSI vacuoles relatively frequently disappear until day 2 compared with IVF inclusions which might be persistent. In addition, spontaneous vacuolization reached a minimum on days 2 and 3 when a presumed ICSI effect would not be suspected any longer. However, a new peak in vacuolization could be noted at compaction stage for both IVF and ICSI embryos. This time vacuoles tended to be higher in number and larger in size (data not shown), probably owing to facilitated fusion after compaction. The present data indicate that de novo formation of vacuoles on day 4 is presumably an embryo-intrinsic phenomenon and closely related to developmental arrest since it is associated not only with a reduced blastulation rate but also with a disastrous blastocyst quality. However, in this study multiple vacuolization events on day 4 were definitely different from early cavitation, which is characterized by clefting of adjacent compacting cells (17). In the extreme, it can become difficult to distinguish between early cavitation and fused vacuoles, as emphasized in the literature (18 20). On the 5th day of development it was clearly recognizable that vacuolization had a negative impact on blastocyst formation (P.0006) at least in ICSI embryos (IVF numbers were too low in this regard). Day 4 turned out to be the most critical day in terms of spontaneous vacuolization. Most interestingly, a statistically significant trend (P.011) toward an elimination of vacuoles from the inner cell mass could be observed because the vast majority of vacuoles could exclusively be located in the trophectoderm (17). Theoretically, this could be seen as a strategy of the embryo to minimize a negative impact of vacuolization on implantation behaviour, though a suspected allocation of euploid cells to the ICM (21), a phenomenon of similar nature, had been doubted (22). It must be admitted that identification of vacuoles at blastocyst stage is a somewhat difficult undertaking. Although scanning of the ICM is rather easy (no vacuoles missed), adequate recording of the trophectoderm turned out to be a challenge owing to the high cell number and small cell size (19). The same situation had to be faced on day 3 of development, when excessive fragmentation and increasing cell number occasionally impaired accurate scoring. Nevertheless, tracing of vacuoles was possible in most cases and exact measuring of the diameter guaranteed detailed information on the further fate of a vacuole whether it disappeared, fused, or split up if situated near a presumed cleavage plane. To summarize, three types of vacuoles could be identified in the present study: (1) those already present at oocyte collection, which may rapidly develop during maturation (3); (2) those artificially created by an embryologist (ICSI); and Fertility and Sterility 1639

6 (3) those accompanied with developmental arrest (ICSI and IVF). Although embryos derived from vacuole-positive predecessors were not transferred according to our laboratory guidelines, they sometimes were the only candidates available (n 3). In those rare cases all patients decided to proceed with transfer. Thus pregnancies (n 2) could be achieved which all resulted in healthy newborns. Similar pregnancy rates in vacuole-positive cycles, e.g., at least one gamete or embryo showing a vacuole (but not transferred), compared with cycles without signs of vacuolization (data not shown) are contradictory to the theory that if one oocyte/embryo is affected the whole cohort is, regardless of the presence of a certain morphological feature (12, 23). To conclude, these data provide new information, e.g., size- and time-dependent negative effect of vacuoles on further development, which may help to inform patients more accurately in terms of cycle prognosis. REFERENCES 1. Van Blerkom J, Antczak M, Schrader R. The developmental potential of the human oocyte is related to the dissolved oxygen content of follicular fluid: association with vascular endothelial growth factor levels and perifollicular blood flow characteristics. Hum Reprod 1997; 12: Mikkelsen AL, Lindenberg S. Morphology of in-vitro matured oocytes: impact on fertility potential and embryo quality. Hum Reprod 2001;16: Van Blerkom J. Occurrence and developmental consequences of abberrant cellular organization in meiotically mature human oocytes after exogeneous ovarian hyperstimulation. J Electron Microsc Tech 1990; 16: Kahraman S, Yakin K, Dönmez E, Samli H, Bahce M, Cengiz C, et al. Relationship between granular cytoplasm of oocytes and pregnancy outcome following intracytoplasmic sperm injection. Hum Reprod 2000;15: Ebner T, Moser M, Sommergruber M, Puchner M, Wiesinger R, Tews G. Developmental competence of oocytes showing increased cytoplasmic viscosity. Hum Reprod 2003;18: De Sutter P, Dozortsev D, Qian C, Dhont M. Oocyte morphology does not correlate with fertilization rate and embryo quality after intracytoplasmic sperm injection. Hum Reprod 1996;11: Balaban B, Urman B, Sertac A, Alatas C, Aksoy S, Mercan R. Oocyte morphology does not affect fertilization rate, embryo quality and implantation rate after intracytoplasmic sperm injection. Hum Reprod 1998;13: Ebner T, Yaman C, Moser M, Sommergruber M, Jesacher K, Tews G. A prospective study on oocyte survival rate after ICSI: influence of injection technique and morphological features. J Assist Reprod Genet 2001;18: Findikli N, Kahraman S, Kumtepe Y, Donmez E, Benkhalifa M, Biricik A, et al. Assessment of DNA fragmentation and aneuploidy on poor quality human embryos. RBM Online 2004; 8: Serhal PF, Ranieri DM, Kinis M, Marchant S, Davies M, Khadum IM. Oocyte morphology predicts outcome of intracytoplasmic sperm injection. Hum Reprod 1997;12: Alikani M, Palermo G, Adler A, Bertoli M, Blake M, Cohen J. Intracytoplasmic sperm injection in dysmorphic human oocytes. Zygote 1995;3: Otsuki J, Okada A, Morimoto K, Nagai Y, Kubo H. The relationship between pregnancy outcome and smooth endoplasmic reticulum clusters in MII human oocytes. Hum Reprod 2004;19: El.Shafie M, Sousa M, Windt ML, Kruger T. An atlas of the ultrastructure of human oocytes. A guide for assisted reproduction. New York: Parthenon, Loutradis D, Drakakis P, Kallianidis K, Milingos S, Dendrinos S, Michalas S. Oocyte morphology correlates with embryo quality and pregnancy rate after intracytoplasmic sperm injection. Fertil Steril 1999;72: Nayudu PL, Lopata A, Jones GM, Gook DA, Bourne HM, Sheater SJ, et al. An analysis of human oocytes and follicles from stimulated cycles: oocyte morphology and associated follicular fluid characteristics. Hum Reprod 1989;4: Van Blerkom J, Bell H, Henry GH. The occurrence, recognition and developmental fate of pseudo-pronuclear eggs after in-vitro fertilization of human oocytes. Hum Reprod 1987;2: Veeck LL, Zaninovic N. An atlas of human blastocysts. New York: Parthenon, Watson AJ. The cell biology of blastocyst development. Mol Reprod Dev 1992;33: Dokras A, Sargent IL, Barlow DH. Human blastocyst grading: an indicator of developmental potential? Hum Reprod 1993;8: Kovacic B, Vlaisavljevic V, Reljic M, Cizek-Sajko M. Developmental capacity of different morphological types of day 5 human morulae and blastocysts. RBM Online 2004;8: Hardy K, Handyside AH, Winston RML. The human blastocyst: cell number, death and allocation during late preimplantation in vitro. Development 1989;107: Evsikov S, Verlinsky Y. Mosaicism in the inner cell mass of human blastocysts. Hum Reprod 1998;13: Meriano JS, Alexis J, Visram-Zaver S, Criz M, Casper RF. Tracking of oocyte dysmorphism for ICSI patients may prove relevant to the outcome in subsequent patient cycles. Hum Reprod 2001;16: Ebner et al. Vacuolization throughout development Vol. 83, No. 6, June 2005