Outcomes of in vitro fertilization with preimplantation genetic diagnosis: an analysis of the United States Assisted Reproductive Technology Surveillance Data, 2011-2012.

A total of 229,096 fresh autologous ART cycles performed in 2011 and 2012 were reported to NASS; among these, 10,407 (4.5%) reported use of PGD. In 2011, 440 (98%) of 451 ART clinics reported at least one PGD cycle; in 2012, all reporting ART clinics (n = 456) performed PGD cycles. After limiting our study to cycles that took place in clinics that performed at least one PGD cycle in 2011 or 2012 and cycles with retrieval of at least one oocyte and transfer at blastocyst stage, the final data set was composed of 106,902 ART cycles, including 9,833 PGD cycles (about 94% of the original 10,407 reported). Among these 9,833 PGD cycles, the majority (55.6%, n = 5,471) were performed for aneuploidy screening, followed by other reasons (29.1%, n = 2,859) and genetic testing (15.3%, n = 1,503) ( Table 1 ). Among cycles where PGD was used for other reasons, only 2% (n = 68) provided further information on specific reasons for PGD use, which mostly included gender selection (n = 63). Table 1 presents patient demographic and clinical characteristics for fresh, nondonor PGD cycles (by PGD reason) and non-PGD cycles. The percentage of all cycles with PGD performed decreased from 4,697 (9.5%) of 49,359 cycles in 2011 to 5,136 (8.9%) of 57,543 cycles in 2012. An increase in PGD use was only detected among cycles with PGD for aneuploidy screening (see Table 1 ). Clinical characteristics, such as pregnancy history, number of prior ART cycles, and factors related to oocyte and embryo quality, varied across the PGD groups. For example, the proportion of women with one or more prior miscarriages was higher in the PGD Aneuploidy group compared with women in the PGD Other group and the PGD Genetic group (53.8% vs. 30.6% and 29.6%, respectively). Of all PGD cycle groups, the PGD Genetic group had the highest number of oocytes retrieved (43.7% had 16 or more oocytes retrieved) and the highest number of embryos transferred (45.1% of cycles resulted in the transfer of ≥2 embryos). Compared with women who did not use PGD, a greater proportion of women who used PGD for the prevention of genetic disorders or for other reasons were younger than 35 years or aged 35–37 years. In contrast, more than half of women (50.9%) undergoing PGD for aneuploidy screening were >37 years, compared with 32.8% of women who did not use PGD. Approximately 74.6% of PGD Genetic and 73.9% of PGD Other cycles resulted in the transfer of one or more embryos as compared with 62.1% of non-PGD cycles; however, the non-PGD group had the highest percentage of embryos cryopreserved. Table 2 presents the odds ratios for the association of age-specific treatment outcomes and use of PGD according to indication after adjusting for confounding factors (infertility diagnosis, pregnancy history, prior ART cycles, and factors related to oocyte and embryo quality). Among cycles with women <35 years involving transfer of at least one embryo, odds of clinical pregnancy and live birth per transfer were lower for all types of PGD cycles than non-PGD cycles. Among live-birth deliveries, those resulting from PGD Genetic cycles had significantly reduced odds than those resulting from non-PGD cycles to be low-birth-weight infants (aOR 0.73; 95% CI, 0.54–0.98). In contrast, PGD Aneuploidy cycles were associated with significantly higher odds of low-birth-weight delivery per live birth compared with cycles without PGD (aOR 1.25; 95% CI, 1.01–1.54). For women 35–37 years, the adjusted odds of clinical pregnancy and live birth tended to be lower for PGD Genetic and PGD Other cycles compared with non-PGD cycles, although most associations were not statistically significant (aOR for PGD Genetic 0.85; 95% CI, 0.65–1.12, and 0.78; 95% CI, 0.59–1.04, respectively; aOR for PGD Other 0.83, 95% CI, 0.69–1.01, and 0.77; 95% CI, 0.63–0.93, respectively). Notably, PGD for aneuploidy screening was statistically significantly negatively associated with miscarriage (aOR 0.62; 95% CI, 0.45–0.87). Although PGD for genetic and other reasons had increased odds for miscarriage, the CI for the association with PGD Genetic included the null value (aOR 1.56; 95% CI, 0.95–2.57, and 1.49; 95% CI, 1.05–2.12, respectively). Among women >37 years of age, PGD Aneuploidy was positively associated with clinical pregnancy (aOR 1.18; 95% CI, 1.05–1.34), live-birth delivery (aOR 1.43; 95% CI, 1.26–1.62), and multiple-birth delivery (aOR 1.98; 95% CI, 1.52–2.57), and negatively associated with miscarriage (aOR 0.55; 95% CI, 0.43–0.70), compared with non-PGD cycles. When we restricted the analysis to clinics with high PGD use (that reported at least 10 IVF cycles and had PGD rates of >25%), the results were not different than those for all clinics with at least one cycle of PGD performed (data now shown).

The findings from this national study of PGD cycles performed in 2011 and 2012 indicate that the most commonly reported indication for PGD was aneuploidy screening, followed by other reasons and the detection of genetic disorders. Among embryo transfer cycles, use of PGD was not associated with improved clinical pregnancy or live-birth rates for women aged <35 years, regardless of the indication. However, PGD for aneuploidy screening was associated with lower odds of miscarriage per pregnancy relative to cycles without PGD among women aged ≥35 years. Furthermore, among women aged >37 years, PGD for aneuploidy screening was associated with a higher likelihood of having a live-birth delivery per transfer, but these live-birth deliveries were also more likely to be multiple births, compared with cycles where PGD was not used. Consistent with other studies ( 15 – 17 ), we found that the majority of women using PGD for the prevention of genetic disorders were younger than 35 years with no prior miscarriages; in contrast, PGD for aneuploidy screening was performed more often among women older than 37 years and those with one or more prior miscarriages. Because the majority of embryos among women older than 37 years are chromosomally abnormal ( 18 , 19 ), it is recommended that these women undergo PGD to screen for aneuploidies and thereby improve pregnancy rates. We found that the percentage of cycles resulting in the transfer of at least one embryo among PGD cycles was higher than that of non-PGD cycles. We also found that fewer embryos were cryopreserved among PGD cycles compared with non-PGD cycles. These findings may reflect the detection and elimination of chromosomally abnormal and aneuploid embryos after PGD, thereby leaving fewer available for cryopreservation. Among women aged <35 years we found the adjusted odds of clinical pregnancy and live birth were lower compared with cycles where PGD was not used, irrespective of the reported indication for PGD. While there is no sufficient evidence to explain the reasons for the lack of beneficial outcomes of PGD among younger women (<35 years), it is possible that this group of women has more complex infertility problems that were developed at earlier age and cannot be easily overcome by the use of PGD. When PGD was performed for aneuploidy screening among women 35–37 years, the rate of miscarriage was significantly lower compared with women of the same age who did not undergo PGD. In addition, among women aged >37 years, PGD for aneuploidy screening was associated with reduced odds of miscarriage and improved likelihood of having a live-birth and a multiple live-birth delivery, thereby suggesting that more viable or euploid embryos were transferred after screening. Thus, PGD for aneuploidy screening may be beneficial in reducing the risk of miscarriage for women aged ≥35 years and for improving the chance of having a live birth in women aged >37 years. Although NASS does not collect information on PGD methods, newer techniques of array comparative genomic hybridization (CGH) were likely used during 2011–2012. As such, our findings are consistent with a number of randomized studies using array CGH, indicating the benefit of aneuploidy screening among women with advanced maternal age ( 20 – 22 ). Additionally, our findings support the counseling criteria and guidelines which recommend the use of PGD for aneuploidy testing among women >37 years as a means of improving their likelihood of ART success ( 14 , 23 ). Our report is among the first to assess PGD use and associated pregnancy outcomes using national data on nearly all ART cycles performed in the United States ( 13 ). In addition, we were able to evaluate a variety of outcomes according to the reason for PGD use among various age strata and limited to blastocyst transfers, thus reducing the potential impact of bias due to patient selection. However, several limitations should be recognized in the interpretation of our findings. Although the NASS collects data on PGD use and reasons for PGD, reporting of reasons for use of PGD may be imprecise and vary by physicians and clinics, particularly among PGD cycles for other reasons. For example, the characteristics of patients undergoing PGD for other reasons vary, making it difficult to interpret the results for this heterogeneous group. The non-PGD group may also have included cycles with known genetic disorders or chromosomal abnormalities that did not undergo PGD for various reasons, such as financial burden and possible false-negative results due to mosaicism ( 24 ). Moreover, NASS does not collect information on whether PGD use is intended for a particular treatment cycle, and some IVF cycles for which PGD is intended may not have embryos available for biopsy. Furthermore, we limited our analysis to cycles where at least one blastocyst embryo was transferred, which represents a selected population of good-prognosis patients. Another important limitation is the lack of information on the embryo morphology in NASS, which does not allow for evaluation or comparison of embryo viability among PGD cycles and non-PGD cycles. Finally, due to the retrospective study design used for our analysis, it is possible that selection bias affected our findings, particularly for the association between PGD for aneuploidy screening and miscarriage, as women with recurrent pregnancy loss may be more likely to undergo PGD. Due to the complexity of PGD techniques, the efficacy of PGD depends on many different factors associated with the patient’s characteristics and embryo’s quality in addition to the type of PGD method used ( 6 ). The fact that NASS does not collect information on biopsy type, the protocol used to select chromosomal abnormalities, embryo-specific morphology or quality, including number of embryos available for biopsy, number of embryos biopsied, number of embryos discarded after PGD (e.g., chromosomally abnormal or aneuploidy embryos), and their diagnostic results, limited our ability to assess the effectiveness of the PGD procedure. Furthermore, comparing pregnancy results from PGD use can be challenging because different PGD methods may yield different results. Whereas the current method of chromosomal screening by array CGH has been found to improve pregnancy rates when used as part of a comprehensive screening program ( 20 – 22 ), older methods such as fluorescent in situ hybridization (FISH) have not been shown to improve outcomes ( 12 ). It is also possible that clinics with high rates of PGD performed had better outcomes than clinics that only performed a few PGD cycles. However, the results of our sensitivity analysis indicated that clinics with high (>25%) PGD rates did not have better treatment outcomes than clinics with lower rates. Although NASS collects information on the number of previous IVF cycles, it does not include information on previous PGD cycles or the availability of euploid embryos, particularly for two consecutive PGD cycles, which has been associated with improved pregnancy and implantation rates ( 25 ). Finally, as with any clinical diagnostic test, misdiagnosis of embryos can occur in PGD because of the technical difficulty of handling delicate cells and the fact that the cells can only be tested once; potential errors can occur and may result in the transfer of chromosomally abnormal or aneuploidy embryos ( 2 , 19 , 25 – 27 ). Using data from a national population-based surveillance system with sufficient numbers to examine treatment outcomes for IVF cycles where PGD use was reported, we did not find the use of PGD to be associated with improved rates of clinical pregnancy or live birth among women aged ≤37 years, irrespective of the indication. However, PGD for aneuploidy screening of embryos improved the likelihood of having a live birth among women >37 years. This improved pregnancy outcomes among women with advanced maternal age is likely due to the enhanced PGD technique of array CGH with 24-chromosome analysis ( 20 – 22 ). Therefore, identifying euploid embryos presents the most effective opportunity for elective single-embryo transfer to achieve optimal live-birth delivery rate, even among women with advanced maternal age. While preimplantation genetic testing can improve outcomes in certain patient populations, particularly those with a previous genetically affected child or a family history of chromosomal abnormality, the potential risks and benefits of the procedure should be considered in an effort to optimize both the safety and effectiveness of IVF treatments ( 26 ). Furthermore, collecting accurate information on PGD indication, PGD methods, and the outcomes of biopsied embryos with morphology information is critical as part of the national ART surveillance to better understand the effectiveness of PGD. Well-designed prospective, randomized studies are needed to effectively evaluate the efficacy of PGD.

In 1992, Congress passed the Fertility Clinic Success Rate and Certification Act (FCSRCA), which requires each medical center in the United States that performs ART procedures to report data to the Centers for Disease Control and Prevention (CDC) on every ART procedure initiated, where ART is defined as any procedure in which oocytes or embryos are handled in the laboratory for the purpose of establishing a pregnancy. All ART data are reported annually to the CDC’s Web-based National ART Surveillance System (NASS) ( 7 , 13 ). The data collected in NASS include patient demographics, medical history, infertility diagnoses, clinical information pertaining to the ART procedure, and information regarding resultant pregnancies. The data file is organized with one record per ART cycle performed. Because nonreporting clinics (7% of clinics in operation in 2012) tend to be smaller and perform fewer cycles, the CDC estimates that NASS contains information on over 95% of all ART procedures performed in the United States ( 13 ). The collection of information for NASS on the use of PGD and the reason for its use started in 2004 and has been revised over time; consistent reporting of these data began after 2010. For the current study, the cycles with use of PGD and the reported reason for use were categorized into three mutually exclusive groups based on the indication for PGD use: [1] PGD for genetic disorders or chromosomal abnormality (PGD Genetic), [2] PGD for aneuploidy screening of the embryos (PGD Aneuploidy), and [3] PGD for other or unknown reasons (PGD Other, including gender preference, history of infertility, elevated follicle-stimulating hormone [FSH] levels, obesity, etc.). We also examined the reported reasons for ART use, which included free-text entries (“other specify”) for reasons for use; in some cases, this information was used to reclassify indication for PGD using a hierarchical system. For example, when cycles for which “aneuploidy screening of the embryos” (PGD Aneuploidy) was reported as the reason for PGD but “genetic disease” was listed as the reason for ART, we reclassified the report for the indication for PGD use to “PGD Genetic.” Similarly, if “recurrent miscarriage” was the reason for ART but “other screening for embryos” was reported as the reason for PGD (PGD Other), we reclassified the PGD indication to “PGD Aneuploidy.” Cycles without reported use of PGD were categorized as non-PGD cycles for the purpose of comparison with PGD cycles. Because information on PGD use is not consistently collected for frozen cycles and PGD is often used for routine screening of donor cycles, which often have different outcomes than fresh autologous cycles, we restricted our study to fresh, autologous ART cycles performed in 2011 and 2012 (the latest data available with consistent PGD reporting information). Because PGD procedures are not offered at all ART clinics, we further limited our study to cycles performed in clinics that reported at least one PGD cycle in either 2011 or 2012. Cycles cancelled before oocyte retrieval were excluded. We further restricted our study to cycles with a blastocyst stage embryo available for transfer because PGD nearly always requires culture of the embryo to blastocyst stage (5–6 days after fertilization) and only 1% of the transfers occurred at the cleavage stage. For cycles with and without use of PGD, we examined the distribution of the following patient characteristics: patient age, infertility diagnosis, number of prior ART cycles, number of prior miscarriages, number of prior pregnancies, number of oocytes retrieved, number of embryos transferred, and number of embryos cryopreserved. Patient age at the time of the ART procedure was grouped into three categories, 37 years. The infertility diagnoses assessed included tubal factor, ovulatory dysfunction, diminished ovarian reserve, endometriosis, uterine factor, male factor, and unexplained factor; because more than one diagnosis could be reported, the diagnosis categories were not mutually exclusive. The number of oocytes retrieved was categorized as 1–10, 11–15, and ≥16, and the number of embryos transferred was categorized as no transfer, 1, and ≥2. The number of embryos cryopreserved was classified as none and ≥1. We used two-tailed Pearson’s chi-square tests to compare the distribution of patient characteristics (demographic and clinic) for PGD cycles, by PGD category, versus cycles without PGD. The treatment outcomes we assessed were rate of clinical pregnancy and live-birth delivery per transfer; rate of miscarriage (pregnancy loss) per pregnancy, and rate of multiple birth delivery, preterm delivery, and low birth weight delivery per live birth. We calculated age-specific rates of these treatment outcomes for each category of PGD reason and for cycles without use of PGD. Multivariable logistic regression models were developed to calculate unadjusted and adjusted odds ratios (aOR) and 95% confidence intervals (CI) for the association between the treatment outcomes and the reason for PGD, stratified by age group; non-PGD cycles were the referent. In addition, a subanalysis was conducted of data from 24 clinics that performed at least 10 IVF cycles and had PGD rates of >25% to test whether these clinics have better treatment outcomes than those of all clinics. Statistical analyses were conducted using SAS, version 9.3 (SAS Institute), and P <.05 was considered statistically significant for all values. The study was approved by the CDC’s institutional review board.