Obstetrics and Gynecology, University of Texas Southwestern Medical Center, Dallas, Texas

Transcription

1 Late stages of embryo progression are a much better predictor of clinical pregnancy than early cleavage in intracytoplasmic sperm injection and in vitro fertilization cycles with blastocyst-stage transfer Khurram S. Rehman, M.D., a,b Orhan Bukulmez, M.D., a,b Martin Langley, B.Sc., a Bruce R. Carr, M.D., b Anna C. Nackley, M.D., a,b Kathleen M. Doody, M.D., a,b and Kevin J. Doody, M.D. a,b a Center for Assisted Reproduction, Bedford; and b Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Texas Southwestern Medical Center, Dallas, Texas Objective: To define and validate metrics of embryo progression and morphology during extended embryo culture and to compare the effects of early cleavage (EC) vs. blastulation stages on clinical pregnancy. Design: Retrospective observational study. Setting: University-affiliated assisted reproduction center. Patient(s): One thousand two hundred ninety-two intracytoplasmic sperm injection and 842 IVF blastocysttransfer cycles. Intervention(s): The embryo progression index (EPI) was calculated as the area under the curve of total cell number (TCN) over time, by using observed TCN for cleavage-stage embryos and estimated blastocyst TCN according to morphology. The EPI from days 1 3 measured early cleavage, and blastulation was assessed by EPI over extended embryo culture. Blastocyst morphology was converted into numerical blastocyst quality scores (BQSs). Receiver operating characteristic curve analysis was used to evaluate predictors for clinical pregnancy. Main Outcome Measure(s): Clinical pregnancy. Result(s): Per-cycle mean EPI and mean BQS for all embryos developing into blastocysts, as well as mean BQS of the transferred embryos, were significant predictors of clinical pregnancy in intracytoplasmic sperm injection and IVF cycles. Mean EPI for days 1 3 did not predict outcome. Conclusion(s): Early cleavage is a putative marker of embryo quality. Late-stage embryo development is more sensitive and specific in predicting clinical pregnancy than is early cleavage, supporting the use of extended embryo culture for embryo selection. The embryo progression index and BQS may also be used for this purpose. (Fertil Steril 2007;87: by American Society for Reproductive Medicine.) Key Words: Blastocyst, morphology, embryo progression, early cleavage, in vitro fertilization, clinical pregnancy Assessment of embryo morphology has been an integral part of the selection of embryos for transfer since the earliest days of development of the assisted reproductive technologies. Many detailed assessment and grading schemes have been devised for each stage of embryo development, including pronuclear-stage zygotes and cleavage-stage embryos on days 2 and 3 after fertilization, and for blastocysts. Assessment of morphology in zygotes at the two-pronuclear (2PN) stage includes measures such as pronuclear alignment and morphology. Edwards and Beard (1, 2) have reviewed the evidence that oocyte and pronuclear polarity Received January 14, 2006; revised November 1, 2006; accepted November 7, Presented at the 61 st Annual Meeting of the American Society for Reproductive Medicine, Montreal, Quebec, Canada, October 15 19, Reprint requests: Khurram S. Rehman, M.D., University to Texas Southwestern Medical Center at Dallas, Department of Obstetrics and Gynecology, Division of Reproductive Endocrinology and Infertility, 5323 Harry Hines Boulevard, #J6.114, Dallas, Texas (FAX: ; ksrehman@gmail.com). are important. For cleavage-stage embryos, morphologygrading schemes usually include assessment of blastomere number, blastomere fragmentation, and more recently, blastomere multinucleation (3 6). Grading of blastocyst morphology was first described by Dokras et al. (7), mainly considering the degree of expansion; however, the morphology-grading scheme developed by Gardner and Schoolcraft (8) has gained the widest acceptance. It incorporates assessment of degree of expansion of the blastocyst as well as visual estimation of cell numbers in the inner cell mass (ICM) and the trophectoderm (TE), as described in more detail in Materials and Methods. This grading scheme has been applied for both research purposes and clinically, in selection of embryos for transfer or cryopreservation. Gardner and Schoolcraft blastocyst morphology (8) has been shown to have predictive value for embryo competence, as assessed by implantation and clinical-pregnancy rates, with significantly higher clinical-pregnancy rates with transfer of two top-scoring blastocysts on day /07/$32.00 Fertility and Sterility Vol. 87, No. 5, May 2007 doi: /j.fertnstert Copyright 2007 American Society for Reproductive Medicine, Published by Elsevier Inc. 1041

2 than when either only one top-scoring blastocyst was available or no top-scoring blastocysts were available for transfer on day 5 (9). Compared with assessment of embryo morphology, embryo progression is a less well-studied concept, reflecting dynamic changes (growth and development) of the embryo over time. Progression thus encompasses information from serial embryo assessments over time, whereas morphology is a snapshot of the embryo at one point in time, probably most crucially on the day of embryo transfer. In contrast to the multiple morphology-grading schemes described in the previous two paragraphs for 2PN zygotes, cleavage-stage embryos, and blastocysts, fewer proposals exist for evaluation of embryo progression. One aspect of embryo progression that has been proposed as a putative early marker of embryo quality and competence is the presence of early cleavage (EC), referring to the timing of the first and second blastomere divisions to reach the two- and four-cell stages. The effects of EC on implantation rates were first described by Edwards et al. (10), who found that human embryos undergoing early first and second cleavage divisions had per-embryo implantation rate of 30%, which was significantly higher than those cleaving later. Since then, several groups have defined scoring systems for selection of cleavage-stage embryos for transfer that incorporate elements of both morphology assessment and assessment of EC stages of embryo progression. This was described for 2PN and two-cell stage transfers by Scott and Smith (11), who scored embryos on the basis of pronuclear and nucleolar alignment and morphology and the appearance of the cytoplasm, with increased scores for embryos that had undergone pretransfer EC. They reported a 28% implantation and a 65% delivery rate with the highest mean embryo scores. Similarly, a strong predictive effect of EC on implantation and pregnancy rates has been reported for single-embryo transfers (SETs) and double-embryo transfers (DETs) by using day 2 cleavage-stage embryos (12, 13). For transfer of day 3 cleavage-stage embryos, a set of morphological and progression characteristics of top-quality embryos has been defined, including EC, with high implantation and pregnancy rates with either SET or DET (14, 15). However, limited data exist regarding the relevance of EC to embryo competency after extended embryo culture. Fenwick et al. (16) found that embryos undergoing early first cleavage within 25 hours after insemination had higher progression to the blastocyst stage. Racowsky et al. (17) found that blastocysts developing from embryos with seven or eight blastomeres on day 3 had higher implantation and clinical-pregnancy potential than did those developing from other cleavage stages. The present study has three aims. First, we set out to define novel quantitative metrics of embryo progression and embryo morphology that are relevant to extended embryo culture with transfer of blastocyst or compacted morulastage embryos. Second, we aimed to validate these measures by investigating whether indices of embryo morphology or progression during extended embryo culture had any predictive value in determining clinical pregnancy in fresh, nondonor oocyte intracytoplasmic sperm injection (ICSI) or IVF cycles. Last, we wished to compare the assessment of embryo progression during the EC stages of embryo development with that during the last stages of embryo progression, encompassing morula formation and blastulation, in regard to predicting clinical pregnancy. MATERIALS AND METHODS We undertook a retrospective analysis of serial embryo morphology observations and clinical pregnancy outcome during ICSI and IVF cycles performed at the Center for Assisted Reproduction, a university-affiliated private assisted-reproduction program. Extended embryo culture with use of sequential culture media has been used universally since January All analyzed cycles took place between April 1998 and November The Institutional Review Board of the University of Texas Southwestern Medical Center determined that this study was exempt from review. A total of 1,292 ICSI and 842 IVF cycles were included in the study. The demographics and clinical characteristics of these cycles are summarized in Table 1. The inclusion criteria were fresh cycles, resulting in transfer of at least one embryo on day 5 or day 6 after fertilization. The majority of transfers occurred on postfertilization day 5, with observation of embryos up to days 6 to 7 after fertilization. Exclusion criteria comprised cycles using frozen embryo transfer, use of donor oocytes or donor embryos, or use of gestational carrier surrogacy. Serial daily embryo observations were recorded in the Electronic Medical Record (EMR) software, DOC version 5 (Digital MD Systems, Bedford, TX). Pregnancy outcomes were recorded in the EMR and also were recorded in an outcomes database. Clinical pregnancy was defined as the presence of at least one gestational sac with positive fetal heart activity, detected by transvaginal ultrasound performed days after oocyte retrieval. Controlled Ovarian Hyperstimulation All of the patients were treated with oral contraceptive pills (Desogen; Organon, West Orange, NJ) on cycle days 2 to 4 and discontinued this treatment after 14 to 28 days of administration. All patients received a recombinant FSH preparation (either Follistim [Organon] or Gonal-F [Serono, Rockland, MA]). Pituitary suppression was achieved either by a conventional GnRH agonist protocol (long or flare) or a GnRH antagonist protocol. The GnRH antagonist protocol consisted of precycle suppression and microdose recombinant hcg treatment during FSH stimulation. We have recently described both GnRH agonist and antagonist protocols in detail (18). Upon confirmation of at least three dominant follicles (mean diameter, 18 mm), recombinant hcg (Ovidrel, Se Rehman et al. Blastulation vs. early cleavage Vol. 87, No. 5, May 2007

3 TABLE 1 Demographics and cycle characteristics for ICSI and IVF cycles. Parameter ICSI IVF No. of cycles 1, Female age (y) BMI (kg/m 2 ) Cycle no Day 2/3 FSH (miu/ml) FSH used (IU) 3, , Peak E 2 (pg/ml) 2, , Oocyte no Metaphase II/ mature oocyte rate (%) Fertilization rate (%) No. of embryos transferred Clinical-pregnancy rate, % (n) 34.1 (440/1,292) 42.9 (361/842) Note: Data are expressed as mean SEM, unless otherwise indicated. BMI body mass index. rono; 250 g SC) or urinary hcg (5,000 IU) was administered, with oocyte retrieval occurring hours later. up to two to four blastocysts, depending on patient age and/or embryo quality. Embryo Culture and Transfer After retrieval (day 0), oocytes were placed in human tubal fluid (HTF; Irvine Scientific, Santa Ana, CA) medium plus 12% (by volume) synthetic serum substitute (SSS; Irvine Scientific) and were cultured for approximately 4 hours. After IVF or ICSI, embryo culture was performed in a humidified incubator with 6% CO 2, 5% O 2, and 89% nitrogen at 37 C (18). Fertilization was evaluated hours after insemination or sperm injection (day 1). From April 1998 to February 1999, embryos were cultured for 48 hours in S1 (IVF Science, Vero Beach, FL) medium, followed by 48 to 72 hours culture in S2 (IVF Science) to day 5 or day 6. In February 1999, a media change was made, and G1.2 (IVF Science) and G2.2 or CCM (IVF Science) replaced S1 and S2, respectively. Fertilized oocytes were cultured in Petri dishes containing 100- L microdrops of S1/G1.2 medium under oil, two embryos per drop, for 48 hours. On day 3, cleavage-stage embryos were transferred to fresh 100- L microdrops of S2/G2.2 or CCM medium under oil and cultured, two embryos per drop, for an additional 48 to 72 hours. Blastocyststage embryos with a well-defined ICM and expanding TE were used for transfer on day 5. All patients received luteal phase support with 50 mg of IM P in oil daily, which was initiated on the day after oocyte retrieval. An abdominal ultrasound (5 MHz) was used for all transfers to assist intrauterine placement of the embryotransfer catheter (Edwards-Wallace catheter; Marlow Technologies, Willoughby, OH). Patients were advised to transfer Gardner and Schoolcraft Blastocyst Morphology Grading Blastocysts were graded according to the morphology criteria of Gardner and Schoolcraft (8). Blastocysts were given a numerical score from 1 to 6 on the basis of their degree of expansion and hatching status, as follows: 1, an early blastocyst with a blastocoel that is less than half of the volume of the embryo; 2, a blastocyst with a blastocoel that is half of or greater than half of the volume of the embryo; 3, a full blastocyst with a blastocoel completely filling the embryo; 4, an expanded blastocyst with a blastocoel volume larger than that of the early embryo, with a thinning zona; 5, a hatching blastocyst with the TE starting to herniate though the zona; and 6, a hatched blastocyst, in which the blastocyst has completely escaped from the zona. The development of the ICM was graded as follows: A, tightly packed, many cells; B, loosely grouped, several cells; and C, very few cells. Last, the TE was assessed as follows: A, many cells forming a cohesive epithelium; B, few cells forming a loose epithelium; or C, very few large cells. The blastocyst illustrated in Figure 1 is a 4AA embryo, that is, fully expanded with good ICM and TE morphology. Blastocyst Quality Scoring We sought to define a numerical blastocyst morphologygrading system based on the criteria established by Gardner and Schoolcraft (8). Of the three components of their morphology-grading scheme, degree of expansion and hatching status was already numerically coded by an integer from 1 to 6, whereas ICM and TE grades are represented by one letter each, with A representing the highest grade. We Fertility and Sterility 1043

4 FIGURE 1 The method of conversion of blastocyst morphology grading into the numerical BQS. Blastocyst quality score is the product of degree of expansion (numbered 1 to 6) and ICM and TE grades, where grade A 3, grade B 2, and grade C 1. The embryo illustrated represents an expanded blastocyst with excellent morphology, with a BQS of 36. coded these letter grades into numeric form in the simplest possible manner, using A 3, B 2, and C 1. To combine these three values to yield a single summary score, we evaluated both additive and multiplicative combinations of the three component numbers. For a 4AA embryo, the additive blastocyst quality score (BQS-A) was equal to 4 3 3, giving a BQS-A of 10, whereas the multiplicative blastocyst quality score (BQS-M) was equal to 4 3 3, giving a BQS-M score of 36. We chose the BQS-M measure over BQS-A because it gave a greater spread of values between lower-quality and higherquality blastocysts. This was reflected in slightly better predictive value for clinical-pregnancy outcome for both IVF and ICSI cycles when using BQS-M over BQS-A (data not shown). In the remainder of this study, BQS refers to the multiplicative version, BQS-M. To restate, BQS is a metric of blastocyst quality that is based on established morphologic criteria, and it is defined as the product of degree of expansion and hatching status and ICM and TE grades, where letter grade A is given the value 3, grade B 2, and grade C 1. A sample BQS calculation for a 4AA-grade embryo is illustrated in Figure 1, with a BQS of 36. Table 2 shows examples of Gardner and Schoolcraft (8) morphology grades with their corresponding numerical conversions into BQS. Of note, different combinations of morphological criteria can yield identical BQS values. Embryo Progression Index Although blastocyst morphology scoring is relatively straightforward and relies on an existing blastocyst grading scheme, to measure embryo progression during extended embryo culture, we needed to create a novel measurement scheme. Figure 2 shows the calculation of our proposed measure. The embryo progression curve illustrated represents the observed or estimated total cell number (TCN; see next section) from serial daily observations of a single developing embryo, plotted against time in days after fertilization, in this example for an embryo reaching an early blastocyst stage (Gardner and Schoolcraft grade 1AA) on day 5 of culture. The embryo progression index, or EPI for short, is calculated as the area under this curve (AUC), by using trapezoi- TABLE 2 Gardner and Schoolcraft blastocyst morphology grades, with corresponding BQS. Morphology BQS Morphology BQS Morphology BQS 1AA 9 3AA 27 5AA 45 1AB/1BA 6 3AB/3BA 18 5AB/5BA 30 1BB 4 3BB 12 5BB 20 1AC/1CA 3 3AC/3CA 9 5AC/5CA 15 1BC/1CB 2 3BC/3CB 6 5BC/5CB 10 1CC 1 3CC 3 5CC 5 2AA 18 4AA 36 6AA 54 2AB/2BA 12 4AB/4BA 24 6AB/6BA 36 2BB 8 4BB 16 6BB 24 2AC/2CA 6 4AC/4CA 12 6AC/6CA 18 2BC/2CB 4 4BC/4CB 8 6BC/6CB 12 2CC 2 4CC 4 6CC Rehman et al. Blastulation vs. early cleavage Vol. 87, No. 5, May 2007

5 FIGURE 2 Calculation of EPI. The embryo progression curve shown represents an early blastocyst reaching an estimated TCN of 48 cells on day 5 after fertilization. Embryo progression index is the AUC of observed or estimated TCN plotted against time in days after fertilization; EPI for days 1 5 is the sum of the areas of the four shaded trapezoids: a, b, c, and d. dal integration. For the sample progression curve shown for the embryo cultured to day 5, the EPI is the sum of the four shaded trapezoidal areas: a, b, c, and d (Fig. 2). The calculation of observed or estimated TCN is described in the following section. Total Cell Number Observation and Estimation Total cell number represents the overall blastomere cell number in the embryo. For cleavage-stage embryos and morulae, TCN is simply the cell count of blastomeres in the embryo. In the blastocyst, TCN represents the sum of the number of cells in the ICM and the number of TE cells. Total cell number is easily assessed noninvasively for cleavagestage embryos by counting the blastomeres using standard inverted light microscopy. Cleavage of the zygote to yield two-cell, four-cell, and eight-cell stages typically results in 4 12 blastomeres by day 3 after fertilization. For cleavagestage embryos, we can thus directly use the cell count as the TCN, with some straightforward examples shown in Table 3. However, for compacting morulae and blastocysts, blastomere boundaries become less well defined, and it often is not possible to accurately count individual blastomeres in the living embryo using standard light microscopy. This precludes rapid visual assessment of TCN. Reliable blastomere counting involves mechanically disrupting the embryo, usually fresh or thawed supernumerary embryos from IVF-ICSI cycles; spreading out the cells; fixing and staining; and then counting the nuclei. Since the original article measuring TCN in human blastocysts by Hardy et al. (19), which found a mean TCN of 58 cells on day 5, there have been a number of published studies in which human blastocysts were fixed and TCN was counted after nuclear staining. The first study by Dumoulin et al. (20) used spare IVF and ICSI embryos that were not frozen and found a mean day 5 blastocyst TCN of The second study from the same group used spare embryos from ICSI only and found a mean day 5 and 6 TCN of between 43 and 53 cells (21). Devreker et al. (22) found that TCN varied between a mean of 62 to 100 on day 6, depending on the amino acid composition of the sequential culture media used. Archer et al. (23) used frozen cleavage-stage embryos that were grown to blastocysts in sequential media after thawing, and they found a mean day 6 TCN of cells. They used two example day 6 blastocysts with corresponding stained and fixed TCN preparations, with a TCN of 42 for an early blastocyst and 106 for an expanded blastocyst. The study by Fong and Bongso (24) represents an outlier, comparing blastocysts grown in sequential media with and without co-culture with Vero cells (African green monkey kidney epithelial cells). Those investigators reported much higher cell numbers than the other studies on day 6, with mean day 6 TCN in good-morphology blastocysts of 64 to 168 cells. In summary, these studies found mean blastocyst TCN of approximately cells on day 5 and 50 to 100 cells on day 6. In the present study, the above data were used to guide the development of a mathematical model to allow estimation of TCN from noninvasive morphology observations conducted on human blastocysts. Using a model to estimate TCN for blastocysts is useful for clinical IVF programs in which destroying the embryo is obviously not an option, and in which safe nonlethal nuclear staining of blastomeres is not currently possible. TABLE 3 Examples of TCN values for cleavage-stage embryos and compacting morulae. Stage TCN Zygote (2PN, etc.) 1 2 cell 2 4 cell 4 5 cell 5 6 cell 6 7 cell 7 8 cell 8 10 cell cell 12 Morula 24 a a Estimated TCN. Fertility and Sterility 1045

6 Any assessment of morphology and individual cell numbers in a compacting morulae is difficult; thus, we estimated the TCN to be 24 cells, based on a peak TCN of around cells on day 3 for embryos developing into compacting morulae on day 4, and an approximate doubling of cell number every 24 hours during the early stages of embryo development. This is shown in Table 3. For early to hatched blastocysts, we developed a simple mathematical model to allow estimation of TCN. In defining this model, it was apparent that we needed to assign higher estimated TCN values to blastocysts that had better morphology and/or higher stages of expansion. Of note, the basis of the letter grades in Gardner and Schoolcraft (8) morphology scores is the visual estimation of cell numbers in the ICM and TE, so we incorporated Gardner blastocyst morphology grades into our modeled TCN estimates. For each ICM or TE letter-grade decrease from AA, the estimated TCN was reduced by 1/12; for example, a 4AB blastocyst was assigned an estimated TCN 11/12 that of an AA blastocyst, and a 4BB blastocyst received 10/12 the estimated TCN of a 4AA blastocyst. Table 4 shows estimated TCN by grade of blastocyst. For early to fully expanded blastocysts, from a 1CC up to 4AA grade, inclusive, the mean TCN for all grades is 50 cells. For all grades of hatching and hatched blastocysts, from a 5CC up to 6AA, inclusive, the mean TCN is 105. Comparing EC with Late (Blastulation) Stages of Development By using our proposed EPI, we can compare the effects on pregnancy outcome of the early stages of embryo progression, reflecting EC, with the later stages of blastulation and expansion of the blastocyst. To assess EC, we calculated the AUC up to day 3, as shown in Figure 3A. To assess blastulation, we calculated the AUC up to day 5. To capture data on embryos that were not transferred on day 5 but cultured for 1 2 more days with the intent of cryopreservation of viable blastocysts on days 5, 6, and 7, we also extended the EPI to days 6 and 7. Figure 3B shows calculation of EPI up to day 6. However, extending EPI to days 6 and 7 created a number of censored observations; for example, a blastocyst with 4AA Gardner morphology (BQS 36) on day 5 may be transferred or cryopreserved on day 5 or cultured to day 6 before cryopreservation, leading to apparently missing embryo observations on day 6 and 7 or day 7 alone. Analysis of these censored data is described in the next section. Because TCN increases in a roughly exponential fashion in the cleavage-stage embryo and early blastocyst, with an approximate doubling time for TCN of between 18 and 24 hours, it can be seen from Figure 3 that the EPI is more strongly influenced by the AUC in the later stages of embryo development, such as days 3 through 5 or 6, than during EC-stage development during days 1 3 after fertilization. To give more equal weight to EC and blastulation stages of development, we calculated EPI measures using the binary (base 2 ) logarithm of observed or estimated TCN rather than linear, untransformed TCN (these results are not shown). However, predictive value for clinical pregnancy, by receiver operating characteristic (ROC) curve analysis, was inferior to that obtained using linear TCN data. This also confirmed that the late stages of embryo progression are more predictive of clinical-pregnancy outcome than EC events. Censored Observations Our center cultures embryos up to a maximum of day 7 for the purpose of blastocyst-stage cryopreservation. Embryos are routinely transferred on day 5 at the blastocyst or compacted morula stages of development, with rare day 6 transfers being performed only when there are not adequate embryos to transfer on day 5, but these are found on day 6. TABLE 4 Total cell number estimates for blastocysts according to morphology grading.* Degree of expansion Gardner and Schoolcraft morphology grading AA AB/BA BB AC/CA BC/CB CC * Reprinted from Bukulmez et al. Precycle administration of GnRH antagonist and microdose HCG decreases clinical pregnancy rates without affecting embryo quality and blastulation. Reprod Biomed Online 2006;13:465 75, with permission from Reproductive Healthcare Ltd Rehman et al. Blastulation vs. early cleavage Vol. 87, No. 5, May 2007

7 FIGURE 3 Use of the EPI to quantify early cleavage vs. blastulation stages of embryo development. The embryo progression curves shown represent TCN plotted on a linear scale against time in days after fertilization. On the left, EPI from days 1 to 3 is shown as the shaded AUC, measuring early cleavage. On the right, EPI from days 1 to 6 is shown as the shaded AUC, measuring blastulation. For embryos that were transferred on day 5, we assigned the same TCN observed on day 5 to day 6 and day 7 when calculating EPI. Similarly, embryos transferred or cryopreserved on day 6 were assigned the last observed TCN score from day 6 on day 7. This allowed for simple computation of EPI involving censored data, without recourse to extrapolating or estimating the degree of further progression that may have occurred in culture, had the embryo not been either transferred or cryopreserved. However, an embryo reaching a high-quality blastocyst stage on day 5 would still have a much higher EPI, reflecting the AUC of the estimated TCN curve, than another embryo reaching the same stage of development on day 6 or day 7. Statistical Analysis Intracytoplasmic sperm injection and IVF cycles were analyzed separately by using SPSS, version 12 (SPSS Inc.; Chicago, IL); thus, the means of each parameter are not compared between ICSI and IVF. For each treatment cycle, means of BQS and EPI were calculated and entered as separate variables. Then, ROC curve analysis was used to compare the predictive ability of mean BQS of the transferred blastocysts, mean BQS of the entire blastocyst cohort, and mean EPI during both the cleavage stage only and throughout extended embryo culture, using the AUC of the ROC curves to quantitatively compare the sensitivity and specificity of the measures of interest. The categorical outcome variable was clinical pregnancy. RESULTS Blastocyst quality score was calculated for both transferred blastocysts and for the entire cohort of embryos reaching the blastocyst stage for all 1,292 ICSI and 842 IVF cycles. Complete daily embryo-progression data were available for 462 ICSI and 315 IVF cycles resulting in blastocyst transfer, and these data were used to calculate EPI on days 1 3 and days 1 7, day 7 being the latest possible day of extended embryo culture before cryopreservation or discarding. Figure 4A shows an ROC curve analysis of the predictive value of mean BQS and mean EPI on clinical-pregnancy rates in ICSI cycles. The reference line, equivalent to tossing a coin to determine pregnancy outcome, is the heavy black diagonal. For ICSI cycles alone, mean BQS for the transferred embryos had an area under the ROC curve of Blastocyst quality score for all embryos developing to the blastocyst stage was the strongest outcome predictor, with an AUC of 0.652, and the mean EPI for all blastocysts had an AUC of Figure 4B shows an ROC curve analysis of the EPI for days 1 to 3 and female age in the prediction of clinical pregnancy in ICSI cycles. The EPI for days 1 to 3 was neither a sensitive nor a specific predictor of clinical pregnancy in ICSI cycles. This ROC curve can be seen to cross the reference line, and the area under the ROC curve is 0.461, compared with an AUC of 0.5 for the reference line. For comparison, using female age to predict clinical preg- Fertility and Sterility 1047

8 FIGURE 4 Receiver operating characteristic curves of mean BQS and mean EPI in the prediction of clinical pregnancy. (A) Receiver operating characteristic curves for BQS and EPI as predictors of clinical pregnancy in ICSI cycles. The embryo progression index reflects all available embryo morphology data (days 1 7). (B) Receiver operating characteristic curves for EPI on days 1 3 and female age as predictors of clinical pregnancy in ICSI cycles. (C) Receiver operating characteristic curves for BQS and EPI as predictors of clinical pregnancy in IVF cycles. Embryo progression index reflects all available embryo morphology data (days 1 7). (D) Receiver operating characteristic curves for EPI on days 1 3 and female age as predictors of clinical pregnancy in IVF cycles. nancy resulted in an ROC curve with an AUC of This indicated some predictive ability of female age for clinical pregnancy; however, age alone was less predictive of clinical pregnancy than was either mean BQS for the transferred embryos or mean BQS for all embryos developing to the blastocyst stage, or the mean EPI for all blastocysts. Figure 4C shows an ROC curve analysis of BQS and EPI for prediction of clinical pregnancy in IVF cycles. For IVF cycles alone, mean BQS for the transferred embryos was the strongest outcome predictor, with an area under the ROC curve of Blastocyst quality scores for all embryos developing to the blastocyst stage had an AUC of 0.645, and the mean EPI for all blastocysts had an AUC of Figure 4D shows the ROC curves of EPI for days 1 to 3 and female age for prediction of clinical pregnancy in IVF cycles. As found for ICSI, EPI for days 1 to 3 was neither a sensitive nor a specific predictor of clinical pregnancy in IVF cycles. This ROC curve crosses the reference line, and the AUC of the ROC curve is 0.488, again below the reference line AUC of 0.5. Using female age to predict clinical pregnancy resulted in an ROC curve AUC of This indicated much better predictive value of female age for clinical 1048 Rehman et al. Blastulation vs. early cleavage Vol. 87, No. 5, May 2007

9 FIGURE 5 Clinical-pregnancy rates for ICSI and IVF cycles, stratified according to mean BQS of the transferred embryos. Pregnancy rates (%) are expressed as mean SEM. For ICSI cycles, implantation rates for EPI of and were 34.8% (168/482) and 17.1% (73/427), respectively (odds ratio: 2.59, 95% confidence interval: , P.001). Implantation rates for the groups with mean BQS transferred of 10.3 and 10.3 were 41.3% (151/365) and 16.3% (84/514), respectively (odds ratio: 3.61; 95% confidence interval: ; P.001). For IVF cycles, implantation rates for EPI of and of were 47.3% (151/319) and 23.2% (71/306), respectively (odds ratio: 2.97; 95% confidence interval: ; P.001). Implantation rates when mean BQS transferred was 8.3 and 8.3 were 47.4% (150/316) and 23.1% (70/292), respectively (odds ratio: 2.87; 95% confidence interval: ; P.001). pregnancy in IVF cycles than we had found for ICSI cycles. However, age alone was less predictive of clinical pregnancy than was either mean BQS for the transferred embryos or for all embryos developing to the blastocyst stage, or the mean EPI for all blastocysts. Clinical-pregnancy rates for ICSI and IVF cycles, stratified according to the mean BQS of the transferred embryos, are provided in Figure 5. All rates are expressed as percentages. For ICSI cycles, mean clinical-pregnancy rate declined from 48.8% 4.5% in the group with the highest mean BQS, equivalent to a 4AA or higher morphology, to 25.6% 2.6% in the group with the lowest mean BQS transferred, equivalent to embryos averaging below 1AA morphology. For IVF, higher pregnancy rates were noted overall compared with ICSI, except in the group with the lowest mean BQS. Again, a sharp decline in mean clinical-pregnancy rate was observed as the mean BQS of the transferred embryos in the group fell, between 60.2% 5.1% in the highest mean BQS group to 25.0% 3.4% in the group with the lowest mean BQS transferred. From the ROC curve analysis predicting clinical pregnancy, the best cutoff values for EPI for days 1 7 of all embryos and mean BQS of the transferred embryos were calculated. For ICSI cycles, EPI (sensitivity, 67.3%; specificity, 64.4%) and BQS of 10.3 (sensitivity, 58.9%; specificity, 67.3%) were the best threshold values in prediction of clinical pregnancy. In IVF cycles, EPI of (sensitivity, 66.7%; specificity, 59.3%) and BQS 8.3 (sensitivity, 66.9%; specificity, 57.7%) were the calculated thresholds. The cycles were then analyzed in two groups in regards to whether EPI or BQS either were greater than or were equal to or below the calculated cutoff values. Implantation rates were calculated for each group as total number of gestational sacs per total number of embryos transferred. DISCUSSION For extended embryo culture with the ultimate aim of blastocyst-stage embryo transfer, to date there have been relatively limited data regarding the effects of blastocyst morphology on clinical pregnancy. Where progression is concerned, there are no published studies regarding measurement of embryo progression over the entire course of extended embryo culture before transfer. The present study addresses the question of which is more important, the morphology of the embryo on the day of transfer, or the journey that the embryo took to progress to that stage. In addition, the relevance of EC to embryo competency has been less well defined for extended culture programs than for cleavage-stage transfers. The abilities of mean BQS of the transferred blastocysts, and mean BQS and EPI days 1 7 of all blastocysts were more powerful for predicting clinical pregnancy than those achieved by mean EPI days 1 3, in both ICSI and IVF cycles (Fig. 4.). This finding suggests that assessment of fertilization on day 1, changing to sequential medium on day 3, and embryo observation and scoring starting on day 5 may be equivalent in value to daily serial embryo observations. Although Gardner and Schoolcraft (8) blastocyst morphology grading is a clinically useful scheme and has been shown to predict implantation rates, there are several limitations associated with the use of an alphanumeric label for each combination of degree of expansion, ICM grade, and TE grade. Some of these limitations, as compared with our proposed numeric BQS, include the inability to calculate the mean blastocyst grade of the transferred embryos where two or more embryos are transferred; the inability to determine the mean grade of the whole embryo cohort at any given point in time, such as on day 5; and the increased difficulty in grouping or stratifying embryos into a number of defined quality levels, for example representing excellent-, good-, fair-, or poor-quality blastocysts. Our data showing clinical-pregnancy rates according to mean BQS of the transferred embryos, almost exclusively representing DETs, are a good example of the greater ana- Fertility and Sterility 1049

10 lytical power that is afforded to assessment of outcome in extended embryo-culture cycles by simply converting blastocyst morphology grades into BQS. Similarly, although EPI is a relatively crude measure of embryo progression during extended embryo culture, it can be calculated by using simple algorithms and easily gathered observations of embryo cell number and blastocyst morphology. Again, having a numeric summary measure facilitates semiquantitative analysis of embryo progression both at early stages of development and later stages of development, encompassing compaction and morula formation, blastocyst formation, and blastocoel expansion. We are thus able to compare the predictive value of EC vs. late-stage embryo development for the first time. Early cleavage has been incorporated into many proposed schemes for selection of cleavage-stage embryos for transfer as a putative marker of embryo quality. In this study, we show that late-stage embryo development is a much more sensitive and specific predictor of pregnancy outcome in extended embryo culture cycles with day 5 transfer. In addition, both EPI over the course of extended embryo culture (days 1 7) and BQS of transferred embryos may also predict implantation rates, as demonstrated. Early-cleavage events are likely to be a reflection of oocyte competence rather than embryo competence, because they occur before embryonic genome activation at the fourto eight-cell stage, approximately 2 3 days after fertilization (25). Several studies have indicated that there is a paternal component to embryo progression to the blastocyst stage (26 28). It is likely that these paternal effects manifest during the post-ec phase. The importance of the late stages of embryo development on days 4 and 5 of extended embryo culture reflects many important events in human embryogenesis occurring after activation of the embryonic genome. These events include polarization of the blastomeres that can be seen with light microscopy and blastomere differentiation, resulting in formation of the ICM and TE. The changes in gene expression occur before visible evidence of polarization. These events, which require both maternal and paternal genetic contribution, may well be more reflective of embryo competency (implantation potential) than earlier events, such as blastomere cleavage on days 2 and 3 and blastomere fragmentation during early embryo development. Our findings are consistent with those of a previous report indicating that progression of supernumerary embryos cultured to the blastocyst stage had a strong predictive effect for implantation and pregnancy outcome in ICSI cycles with cleavage-stage transfer (29). The potential exists for more detailed assessment of events that occur after embryonic genome activation, including assessment of blastomere metabolism, production of soluble or cell-surface markers of embryo competency, and ultimately assessment of blastocyst gene expression. Such research strategies may well lead to advances in our ability to select the blastocysts with the highest intrinsic potential from a cohort of developing embryos. Thus, the aim will be to optimize pregnancy outcome, with the trend toward transfer of fewer embryos. Our data clearly support the use of extended embryo culture for embryo selection. We have shown a significant predictive ability of blastocyst morphology on clinicalpregnancy rates, a much closer surrogate for live-birth rates than assessment of implantation rates alone. In this study, our ability to predict clinical-pregnancy outcome for blastocyst-stage transfer cycles on the basis of assessment of either blastocyst morphology or embryo progression to the blastocyst stage compares favorably with reported studies on the predictive ability of morphology assessments of cleavage-stage embryos with clinical-pregnancy outcome of cleavage-stage transfer cycles. Our data also compare well with studies using detailed morphological assessment of the pronuclear-stage zygote. Our results for mostly DET, showing that high clinicalpregnancy rates can be achieved with low numbers of embryos transferred, supports the concept of using extended embryo culture and blastocyst transfer for selection of embryos for elective SET. However, further studies, ideally of a prospective nature, need to be performed for elective SET, both to confirm these findings for SET and to provide more detailed outcome information for clinicians and patients contemplating elective SET as a means to address the high multiple-pregnancy rates of many assisted reproductive technology programs. We expect that SET cycles will provide detailed per-embryo outcome data, potentially allowing further refinement of the assessment of both blastocyst morphology and blastocyst implantation and pregnancy potential. The major criticism of the EPI as described here is that it does not incorporate observations of either the zygote or of properties of cleavage-stage embryos other than blastomere number, which have been shown to affect embryo competence and implantation rates, particularly in programs with cleavage-stage transfer that do not practice universal or selective extended embryo culture. These parameters include pronuclear and polar-body morphology, blastomere fragmentation, and multinucleation. High fragmentation in day 3 embryos has been reported to result in lower blastulation rates during extended culture, compared with embryos with 20% fragmentation (30). There is also some evidence suggesting that blastomere fragmentation during the cleavage stages may be reflected later in reduced cell numbers in the TE, and possibly in the ICM (31). This would obviously impact both blastocyst morphology grading and our estimated TCN numbers, which would thus lower BQS and EPI, respectively. Other parameters of importance in the cleavage stages of development, such as multinucleation, are not presently in Rehman et al. Blastulation vs. early cleavage Vol. 87, No. 5, May 2007

11 corporated into the proposed BQS and EPI measures. Kovacic et al. (32) have reported that detailed assessment of blastocyst morphology incorporating multinucleation and fragmentation has very high correlation with implantation and clinical-pregnancy rates. In addition, more detailed quantitative morphologic assessment of the ICM has been shown to be more predictive than assessment of TE cell number (33). These studies highlight the need for further investigation to optimize our ability to select the embryos with the highest developmental competency for blastocyststage transfer. Our current state of knowledge of blastocyst morphology will likely be extended and refined as worldwide experience with extended embryo culture continues to progress. Nevertheless, we have described practical, semiquantitative numeric summary measures (metrics) of blastocyst morphology and embryo progression, which are of particular interest to extended embryo culture programs. These measures, EPI and BQS, may be applied in a variety of clinical situations and in comparisons of different embryo culture systems. Possible applications range from comparing the effects of use of different culture system components such as sequential culture media, protein sources, or different physical culture conditions on embryo progression and morphology, to clinical comparisons such as prospective or retrospective studies of the effects of different down-regulation or stimulation protocols on embryo progression and morphology. Further clinical evaluation of these measures is warranted, ideally in a prospective fashion. REFERENCES 1. Edwards RG, Beard HK. Oocyte polarity and cell determination in early mammalian embryos. Mol Hum Reprod 1997;3: Edwards RG, Beard HK. Is the success of human IVF more a matter of genetics and evolution than growing blastocysts? Hum Reprod 1999; 14: Puissant F, Van Rysselberge M, Barlow P, Deweze J, Leroy F. Embryo scoring as a prognostic tool in IVF treatment. Hum Reprod 1987;2: Bolton VN, Hawes SM, Taylor CT, Parsons JH. Development of spare human preimplantation embryos in vitro: an analysis of the correlations among gross morphology, cleavage rates, and development to the blastocyst. J In Vitro Fert Embryo Transf 1989;6: Steer CV, Mills CL, Tan SL, Campbell S, Edwards RG. The cumulative embryo score: a predictive embryo scoring technique to select the optimal number of embryos to transfer in an in-vitro fertilization and embryo transfer programme. Hum Reprod 1992;7: Pickering SJ, Taylor A, Johnson MH, Braude PR. An analysis of multinucleated blastomere formation in human embryos. Hum Reprod 1995;10: Dokras A, Sargent IL, Barlow DH. Human blastocyst grading: an indicator of developmental potential? Hum Reprod 1993;8: Gardner DK, Schoolcraft WB. In vitro culture of human blastocysts. In: Jansen R, Mortimer D, ed. Towards reproductive certainty: fertility and genetics beyond 1999: the plenary proceedings of the 11th World Congress on In Vitro Fertilization & Human Reproductive Genetics. Pearl River, NY: Parthenon, 1999: Gardner DK, Lane M, Stevens J, Schlenker T, Schoolcraft WB. Blastocyst score affects implantation and pregnancy outcome: towards a single blastocyst transfer. Fertil Steril 2000;73: 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 Transfer 1984;1: Scott LA, Smith S. The successful use of pronuclear embryo transfers the day following oocyte retrieval. Hum Reprod 1998;13: Salumets A, Hyden-Granskog C, Makinen S, Suikkari AM, Tiitinen A, Tuuri T. Early cleavage predicts the viability of human embryos in elective single embryo transfer procedures. Hum Reprod 2003;18: Van Montfoort AP, Dumoulin JC, Kester AD, Evers JL. Early cleavage is a valuable addition to existing embryo selection parameters: a study using single embryo transfers. Hum Reprod 2004;19: Gerris J, De Neubourg D, Mangelschots K, Van Royen E, Van de Meerssche M, Valkenburg M. Prevention of twin pregnancy after in-vitro fertilization or intracytoplasmic sperm injection based on strict embryo criteria: a prospective randomized clinical trial. Hum Reprod 1999;14: Van Royen E, Mangelschots K, De Neubourg D, Valkenburg M, Van de Meerssche M, Ryckaert G, et al. Characterization of a top quality embryo, a step towards single-embryo transfer. Hum Reprod 1999;14: Fenwick J, Platteau P, Murdoch AP, Herbert M. Time from insemination to first cleavage predicts developmental competence of human preimplantation embryos in vitro. Hum Reprod 2002;17: Racowsky C, Combelles CM, Nureddin A, Pan Y, Finn A, Miles L, et al. Day 3 and day 5 morphological predictors of embryo viability. Reprod Biomed Online 2003;6: Bukulmez O, Rehman KS, Langley M, Carr BR, Nackley AC, Doody KM, et al. Precycle administration of GnRH antagonist and microdose HCG decreases clinical pregnancy rates without affecting embryo quality and blastulation. Reprod Biomed Online 2006;13: Hardy K, Handyside AH, Winston RM. The human blastocyst: cell number, death and allocation during late preimplantation development in vitro. Development 1989;107: Dumoulin JC, Coonen E, Bras M, van Wissen LC, Ignoul-Vanvuchelen R, Bergers-Jansen JM, et al. Comparison of in-vitro development of embryos originating from either conventional in-vitro fertilization or intracytoplasmic sperm injection. Hum Reprod 2000;15: Dumoulin JM, Coonen E, Bras M, Bergers-Janssen JM, Ignoul-Vanvuchelen RC, van Wissen LC, et al. Embryo development and chromosomal anomalies after ICSI: effect of the injection procedure. Hum Reprod 2001;16: Devreker F, Hardy K, Van den Bergh M, Vannin AS, Emiliani S, Englert Y. Amino acids promote human blastocyst development in vitro. Hum Reprod 2001;16: Archer J, Gook DA, Edgar DH. Blastocyst formation and cell numbers in human frozen-thawed embryos following extended culture. Hum Reprod 2003;18: Fong CY, Bongso A. Comparison of human blastulation rates and total cell number in sequential culture media with and without co-culture. Hum Reprod 1999;14: Braude P, Bolton V, Moore S. Human gene expression first occurs between the four- and eight-cell stages of preimplantation development. Nature 1988;332: Janny L, Menezo YJ. Evidence for a strong paternal effect on human preimplantation embryo development and blastocyst formation. Mol Reprod Dev 1994;38: Shoukir Y, Chardonnens D, Campana A, Sakkas D. Blastocyst development from supernumerary embryos after intracytoplasmic sperm injection: a paternal influence? Hum Reprod 1998;13: Seli E, Gardner DK, Schoolcraft WB, Moffatt O, Sakkas D. Extent of nuclear DNA damage in ejaculated spermatozoa impacts on blastocyst development after in vitro fertilization. Fertil Steril 2004;82: Balaban B, Urman B, Sertac A, Alatas C, Aksoy S, Nuhoglu A. Progression of excess embryos to the blastocyst stage predicts pregnancy and implantation rates after intracytoplasmic sperm injection. Hum Reprod 1998;13: Fertility and Sterility 1051