Severity of Diminished Ovarian Reserve and Chance of Success with Assisted Reproductive Technology.

Among 504,266 IVF cycles for women aged <41, the cancellation rate per cycle increased with increasing maximum serum FSH from 4.9% for FSH 0–5 mlU/mL to 18.9% for FSH ≥21 mlU/mL ( Figure 1A ). For women 21 mIU/mL ( Figure 1B ). Among fresh autologous IVF cycles resulting in retrieval for women aged <41 (n=469,895), 5.8% of cycles (n=27,484) were characterized as severe DOR, 15.2% (n=71,410) as moderate DOR, and 79% (n=371,001) as no DOR ( Table I ). A higher percentage of older women were in the severe and moderate DOR groups than in the no DOR group. Race and obstetric history were similarly distributed among the 3 groups. BMI values in the overweight and obese ranges were more common in the no DOR group. A concomitant infertility diagnosis of ovulatory dysfunction was less common in the severe and moderate DOR groups versus the no DOR group (4.22% and 5.51% vs. 18.48%). Prior ART with no previous live births was more common in the severe and moderate DOR groups. Cancellation of cycles between retrieval and transfer, use of natural cycle protocol, retrieval of 0–4 oocytes, use of ICSI, use of assisted hatching, transfer of 0–1 embryos, transfer of cleavage stage embryo, and having zero embryos available for cryopreservation were more common in the severe DOR group than in the moderate or no DOR groups. Among women <35 years old, severe DOR was associated with a significantly higher chance of cancellation after cycle start as compared to no DOR (37.7% vs. 5.8%, aRR=6.4, CI 6.2–6.6) ( Table II ). Among women <35 years old who made it to transfer, absolute pregnancy and live birth rates were lower in those with severe DOR than in those without DOR; upper confidence limits of adjusted relative risk estimates approached the null value (aRR, 95% CI=0.94, 0.91–0.98 and 0.95, CI 0.91–0.99 for pregnancy and live birth, respectively). A similar pattern was noted for moderate DOR, as well as for ages 35–37 and 38–40 years, although among the older age groups fewer differences were statistically significant. In all age groups, no significant relationship was noted between diagnosis and miscarriage. Among women in all age groups with severe DOR who underwent embryo transfer, traditional stimulation (as compared to mild stimulation or natural cycle) was associated with a higher absolute percentage of pregnancy and live birth; however, after adjusting for cycle characteristics, the only significant difference between different stimulation types was noted for women ages 38–40, for whom natural cycle was associated with a significantly increased likelihood of pregnancy (aRR=1.68, CI 1.37–2.07), but not live birth (aRR=1.35, CI 0.99–1.83) ( Table III ). Among women with moderate DOR, mild stimulation was associated with a lower chance of live birth for women <35 years of age (aRR=0.84,95% CI 0.71–0.99), and lower chance of pregnancy (aRR=0.86, 95% CI 0.75–0.97) and live birth (aRR=0.84, 95% CI 0.71–0.99) for women aged 38–40. In a subgroup sensitivity analysis with more stringent definitions for stimulation type, the findings were similar—traditional stimulation was associated with a higher absolute chance of pregnancy and live birth per transfer; however, after adjusting for cycle characteristics, no significant differences in pregnancy or live birth were found between different stimulation types in any age group. Additionally, a similar pattern was noted when results were tabulated per retrieval rather than per transfer.

Study data were obtained from NASS, a federally mandated reporting system which collects patient demographics, medical and obstetric history, infertility diagnoses, detailed parameters of each treatment cycle, and, if applicable, the resultant pregnancy outcome for all ART cycles performed in the United States (Fertility Clinic Success Rate and Certification Act of 1992 [FCSRCA], Public Law No. 102–493, October 24, 1992). 10 NASS, for which annual validation of 7–10% randomly selected reporting clinics is performed, is estimated to include data from >97% of all ART cycles completed in the United States. 10 Among women <41 years old we calculated cancellation rate per cycle by maximum serum FSH level (n=504,266). We also calculated pregnancy rate per transfer by maximum serum FSH, stratified by patient age. Maximum serum FSH is defined as the maximum early follicular phase FSH value to date for an individual patient. We used linear regression models to evaluate trends in the association between serum FSH level and cancellation and pregnancy rates. Cancellation is defined as any cycle for which stimulation is initiated but for which a transfer does not occur; cancellation can occur prior to retrieval or between retrieval and transfer. For the subsequent analyses our study population was derived from all reported fresh autologous IVF cycles resulting in retrieval for women <41 years (n=688,257) between 2004 and 2012. During those study years NASS defined DOR as “reduced fecundity related to diminished ovarian function, including high FSH or high estradiol measured in the early follicular phase or during a clomiphene challenge test, reduced ovarian volume related to congenital, medical, surgical or other causes, or advanced maternal age (>40).” We excluded all cycles in which the female partner was ≥41, as some women without true clinical DOR may have been grouped as “DOR” based on age alone rather than based on a clinical indicator such as low AMH, elevated serum FSH, or low antral follicle count. We divided the remaining cycles into groups using a hierarchical classification where diagnosis of DOR was the primary consideration. In the absence of a diagnosis, a combination of oocyte yield and maximum serum FSH were used to assign groups as follows: “No DOR” was defined as no diagnosis of DOR and serum FSH 10 mIU/mL and 5–9 oocytes retrieved; and “Severe DOR” was defined as diagnosis of DOR and serum FSH 15 mIU/mL or greater, or diagnosis of DOR and <5 oocytes retrieved, or no diagnosis of DOR and <5 oocytes retrieved and serum FSH ≥15 mIU/mL. We also divided the cycles by stimulation type, specifically “natural cycle” defined as no stimulation medication, “mild stimulation” defined as oral medication such as clomiphene citrate combined with gonadotropins, and “traditional stimulation” defined as gonadotropins only for stimulation combined with either standard GnRH agonist downregulation, GnRH agonist flare, or GnRH antagonist. We performed a sensitivity analysis with a narrower definition for the stimulation groups: “mild stimulation” defined as oral medication plus 0–1,500 IU of FSH and “full stimulation” as no oral medication plus ≥ 1,501 IU of exogenous FSH medication. After excluding 218,362 cycles that could not be accurately classified into DOR categories, largely due to missing (176,772 cycles) or out-of-range values for FSH (462 cycles with value 200), our study population included 469,895 cycles. We compared the distribution of patient and clinical characteristics between severe, moderate, and no DOR groups using Pearson’s χ 2 tests. The primary outcomes of interest were clinical pregnancy, live birth at ≥20 weeks gestational age, and miscarriage at <20 weeks gestational age. Bivariate analyses were conducted to explore the relationship between severity of DOR as compared with no DOR, to explore those outcomes, and to assess possible effects of stimulation type (natural cycle and mild stimulation as compared with traditional stimulation) on outcomes among women with moderate or severe DOR. For both analyses we stratified cycles by maternal age at time of oocyte retrieval (<35, 35–37, and 38–40 years). We used multivariable log binomial regression, stratified by female age (<35, 35–37, 38–40 years) to compare rates of cancellation, pregnancy, live birth, and miscarriage among women with severe or moderate DOR to those with no DOR, and to assess effects of stimulation type on those outcomes, while adjusting for the following maternal and treatment characteristics: infertility diagnosis, cycle history, prior pregnancies, number of oocytes retrieved, use of intracytoplasmic sperm injection (ICSI), use of assisted hatching, number of embryos transferred, embryo stage, and number of embryos cryopreserved; unadjusted and adjusted relative risks with 95% confidence intervals are reported. Calculating outcomes per transfer allowed for adjustment for cycle characteristics such as use of assisted hatching, number of embryos transferred, embryo stage, and number of embryos cryopreserved that would otherwise have not been possible had outcomes been calculated per cycle start or per retrieval. Statistical significance was determined using an alpha of 0.05. All analyses were conducted using SAS v. 9.3. This study was approved by the Institutional Review Board of the Centers for Disease Control and Prevention.

Although increasing maximum serum FSH is associated with increasing risk of cancellation, an FSH threshold was not found at which pregnancy rate approaches zero. These results are in contrast to several smaller studies that have suggested serum FSH >14 mIU/mL or AMH <0.2 ng/mL is associated with a nearly 0% chance of live birth after IVF. 3 , 7 , 8 Overall, our results show that, among women who avoid cancellation and make it to transfer, only minimal differences in the likelihood of live birth exist between women with and without DOR, regardless of severity. While cancellation rate was higher and absolute chance of pregnancy was lower in the severe DOR group, adjusted rate ratio estimates for pregnancy, live birth, and miscarriage outcomes overall suggest little difference between women with and without severe DOR who undergo embryo transfer. This supports prior studies that did not find a significant association between elevated FSH and decreased pregnancy rate 4 – 6 and suggests that those who found elevated FSH to be predictive of decreased pregnancy rate and increased aneuploidy and miscarriage risk 1 – 3 may not have controlled for confounding factors such as age. The large sample size and inclusion of all reporting U.S. clinics allowed for an analysis of a larger portion of young women with severe DOR and, as a result, an ability to investigate outcomes specifically among young women with severe DOR. Among women of all ages <41 who proceed with autologous IVF and reach embryo transfer, traditional stimulation (as compared to mild stimulation or natural cycle) was associated with a higher absolute percentage of cycles resulting in live birth; however, after adjusting for cycle characteristics, no significant association was noted between stimulation type and live birth among women with severe DOR. The absolute pregnancy rates among natural cycles may initially appear higher than one would expect; however, the rates are reported per transfer rather than per cycle start. Among patients with severe DOR able to successfully complete an embryo transfer, the stimulation protocols were not associated with a lesser or greater chance of pregnancy overall. Admittedly, our data lack detail regarding nuance of stimulation protocols such as use of adjunct medications like growth hormone and dehydroepiandrosterone (DHEAS). However, it is somewhat reassuring that no one class of protocols consistently demonstrates clear advantage over another as long as the patient is able to reach embryo transfer. Limitations of our study include cycle- rather than patient-based data collection, lack of embryo quality data, exclusion of a large proportion of data as a result of our classification method and limited available cycle characteristics that pertain to DOR, and the retrospective observational nature of the study. To minimize the effects of these limitations, we were able to control for number of prior failed IVF cycles and for number of supernumerary embryos cryopreserved, which has been shown to correlate with embryo quality. 11 Additionally, randomly selected clinics undergo annual validation that has shown overall discrepancy rates <5%. Admittedly, the DOR discrepancy rate was 6.5% in 2012 10 ; part of the rationale for our defining characteristics was to more accurately capture patients’ ovarian reserve. Some of the classification limitations have been described previously 12 and may be improved in the future; while the current NASS collection system lacks characteristics such anti-Müllerian hormone level and considers age >40 to be diagnostic of DOR, revisions to the collection questionnaire that address these weaknesses are in progress. Strengths of our study include its large sample size and generalizability resultant from its inclusion of clinics throughout the United States, the proven validity of the reported data, and the ability to control for demographic and cycle characteristics, namely age, concomitant infertility diagnosis, ART cycle history, obstetric history, number of oocytes retrieved, use of ICSI, use of assisted hatching, number of embryos transferred, embryo stage, and number of embryos cryopreserved. Providing the option of IVF using autologous oocytes to young women with severe DOR who desire a biologically related child is likely warranted; while risk of cancellation increases with increasing serum FSH, no clear serum FSH threshold was found beyond which the chance of pregnancy with IVF approaches zero, suggesting that ART has potential for success even among women with severe DOR. In this population no particular ovulation protocol was consistently associated with improved outcomes. Overall, our results indicate that the chance of pregnancy among young women with severe DOR who are able to complete embryo transfer is comparable to that of women of the same age and with similar characteristics but without diminished reserve. Moreover, once pregnancy was achieved, women with DOR were not at increased risk of miscarriage as compared to women of similar age without DOR.