Background

Clinical studies have indicated that smooth endoplasmic reticulum aggregation (SERa or SER+) may increase the risk of birth malformation or other abnormal outcomes [1,2,3,4,5], miscarriage [2], and perinatal complications [3]. However, SER+ is quite common. During in vitro fertilization (IVF), SER+ appears in 10% of ovulation-induction cycles and in 19–34% of oocytes [6]. It is recommended to avoid SER+ embryos entirely [7, 8] and advised to also measure SER+ size [2]. Embryologic research has revealed that the smooth endoplasmic reticulum (SER) regulates early embryonic development via energy accumulation [4] and plays a key role in calcium storage and release [9]. SER+ also augments the oocyte maturation rate [10] and diminishes the fertilization rate [11]. Researchers have even reported the occurrence of complex chromosomal rearrangements with consistent 2q31 deletions [5].

Two systematic reviews undertaken in 2014 and 2019 concluded that SER+ embryos can be used when embryos of sufficient quality are not available [6, 9]. Some studies have suggested that SER+ embryos can develop into healthy infants after embryo transfer [10, 12,13,14,15]. Despite this, another study implied that 25% of IVF centers discard SER+ oocytes prior to intracytoplasmic sperm injection (ICSI) [16]. Thus, we performed a meta-analysis to study the effect of SER+ on live births, birth weight, and the number of metaphase-II (MII) oocytes per cycle.

Methods

Database search and screening studies

Two authors (HQZ and WHH) independently conducted a literature search using PubMed (Medline), ScienceDirect, Cochrane, Embase, Ovid, and Scopus. Keywords used to search PubMed were endoplasmic reticulum AND (aggregation OR aggregate OR aggregates OR cluster OR clusters) AND (oocyte OR oocytes OR zygote OR zygotes OR embryo OR embryos) AND (“1978/01/01”[PDAT]: “2020/08/31”[PDAT]). The two authors then independently decided whether an article was to be assessed or not according to our study-eligibility criteria shown in Fig. 1 and Table 1. We did not specify randomized clinical trials (RCTs), as most of the included literature did not mention them.

Fig. 1
figure 1

Summary of study selection. A total of 1500 studies were searched, 997 duplicate articles were removed, and 117 reports that were not related were excluded according to the title and abstract. After reading 386 studies, two additional articles were uncovered from the references. Finally, 11 studies were selected

Full size image
Table 1 Inclusion and exclusion criteria
Full size table

Data extraction

Data extraction and assessment of risk bias were performed according to both Sargeant [17] and O’Connor [18]. Briefly, after appropriate data extraction, another author re-evaluated the details. When there was disagreement, a third author (YZ) was brought in to help establish a consensus. Potential bias was assessed by visual inspection of Begg’s funnel plots, and by using Egger’s linear regression [19] and Begg’s rank correlation tests. We performed statistical analyses using Stata 12.0 (Stata Corp, College Station, TX, USA), and p < 0.05 was considered to be statistically significant. A SER+ cycle was designated as 1 cycle in which one or more oocytes were SER+. We defined a SER+ MII oocyte as an oocyte that was observed to be SER+ and that developed to the MII stage [13].

Meta-analysis

According to the type of raw data extracted, we used a continuous method to calculate the SER+ MII oocyte rate and infant body weight at birth. Effects of SER+ cycles and of SER+ MII oocytes on the number of births were analyzed with a dichotomous method. We assessed the SER+ cycle effect on births, specifically calculating births from MII oocyte number and embryo transfer (ET) number. To understand how SER+ and ovum pick up (OPU) number were related, we assessed the number of MII oocytes per cycle and noted the effect of SER+ cycles on birth weight. Heterogeneity was defined using a Higgins statistic, a p-value, and an I2 statistic (I2 > 50% indicated high heterogeneity) in a previous meta-analysis [20]. We used a fixed-effects analysis in the absence of heterogeneity and a random-effects analysis in the presence of heterogeneity, with subgroups in meta-analysis.

Births were defined as the number of new births divided by the number of ETs. SER+ cycles included at least one MII oocyte [13]. We carried out data analysis using Review Manager (Copenhagen: Nordic Cochrane Centre, Cochrane Collaboration, Version 5.4) for meta-analysis.

Results

We ultimately included in our meta-analysis 11 eligible studies that were published less than 12 years ago (Table 2) [3, 4, 10, 12,13,14,15, 21,22,23,24], and we noted that ovarian stimulation protocols were very common among them. We analyzed births and SER+ cycles (Fig. 2a) and SER MII oocytes (Fig. 2b), and observed no differences between SER+ (OR = 1.13, 95% CI = 0.99–1.3; p = 0.35) and SER− cycle/MII oocytes (OR = 0.98, 95% CI = 0.64–1.5; p = 0.13). Results were also similar with respect to birth weights (Fig. 3a and b). We uncovered no association between SER+ cycle/MII oocytes (OR = -1.71, 95% CI = -87.98–84.55; p = 0.33) and infant birth weight (OR = 159.86, 95% CI = -41.36–361.36; p = 0.67). Figure 4a shows that SER+ cycles generated more MII oocytes, and we posit that hormonally stimulated ovaries appear to produce too many oocytes per cycle and induce SER+ (OR = 1.28, 95% CI = 0.41–2.15; p = 0.004). SER+ also had a positive correlation with oocyte maturation rate (OR = 1.27, 95% CI = 1.1–1.47; p = 0.002; Fig. 4b). The heterogeneity we observed appeared to be due to maternal age (Fig. 4a and b).

Table 2 Characteristics of studies included in the review
Full size table
Fig. 2
figure 2

Forest plot of SER and births. a. Effect of SER+ cycles on births. Analysis of SER+ cycle and SER− cycle groups. There was no effect on birth. b. SER MII oocyte effect on births. Analysis of SER+ MII oocytes and SER− MII oocyte groups. There was no effect on birth. Total means of ET numbers, with the calculation based upon ET embryo number. CI = 95% confidence interval

Full size image
Fig. 3
figure 3

Forest plot of SER effect on birth weight. a. SER cycle effect on birth weight. Analysis of SER+ cycle and SER− cycle groups. There was no effect on birth weight. b. SER MII oocyte effect on birth weight. Analysis of SER+ MII oocyte and SER− MII oocyte groups. There was no effect on birth weight. Total indicates the birth number. CI = 95% confidence interval

Full size image
Fig. 4
figure 4

Forest plot of effect of MII oocytes per cycle on SER. a. Analysis of SER+ cycle and SER− cycle groups. SER+ cycles produced more MII oocytes. Stimulated ovaries producing too many oocytes per cycle were positively correlated with SER+. b. Analysis of SER+ MII oocyte-maturation rate. SER+ was positively correlated with oocyte maturation rate. CI = 95% confidence interval

Full size image

Our data indicated that SER+ cycles produced more oocytes but achieved the same number of births after ET, so many oocytes were wasted. Thus, we defined MII-oocyte use as the number of new births divided by the number of MII oocytes inseminated and compared SER+ and SER− cycles. Figure 5 shows that the SER+ cycle (2.51%, 316/12,578) data were not statistically different from SER− cycle groups (2.72%, 2721/99,935) according to MII-oocyte use. We did, however, observe heterogeneity due to using different embryonic culture incubators. Supplemental Fig. 1 shows funnel plots of MII-oocyte use relative to birth; there was no potential bias, and this lack of potential bias was also corroborated by Egger’s (P = 0.263) and Begg’s tests (Pr > |z| = 1).

Fig. 5
figure 5

Forest plot of MII-oocyte use. Total indicates the number of MII oocytes. Analysis of SER+ cycle and SER− cycle groups. No statistical differences were observed between groups, with the calculation based upon MII-oocyte number. CI = 95% confidence interval

Full size image

Discussion

Primary finding

SER+ did not affect birth or infant body weight. Although ovarian stimulation protocols produced an overabundance of oocytes per cycle, and in our study caused SER+, we interestingly retrieved more MII oocytes from SER+ cycles, and SER+ was positively correlated with oocyte maturation rate.

In normal oocytes, SER manifests as multiple scattered spherical aggregates surrounded by mitochondria [25], whereas the phenomenon of SER+ shows aggregations in a single pattern [26]. SER+ may exist normally in MII oocytes and then disappear prior to pronuclear formation [2], and SER+ may also be present in unfertilized oocytes, degenerated oocytes, and embryos [21].

How SER+ occurs remains unclear, although studies suggest that it may originate from genetic abnormalities. SER+ can occur repeatedly in multiple ICSI cycles in the same patient, which has been explained by genetic factors [11, 27]. Intriguingly, studies have shown that SER+ is associated with elevated anti-Müllerian hormone (AMH) [28]. SER+ in the cytoplasm of oocytes is also correlated with E2 on the hCG-treatment day: the higher the peripheral blood E2 concentration, the more likely it is for SER+ to materialize [28]. Oocyte SER+ may also be caused by overstimulation of the ovaries [29]; for example, increasing GnRH and prolonging the duration of ovarian stimulation increases SER+ [3, 6, 30]. Maternal age and FSH dose are unrelated to the appearance of SER+, although follicle stimulation and oocyte collection significantly increase its risk [10]. Thus, the collection of numerous oocytes in 1 cycle might induce SER+.

Some investigators have reported negative biologic impacts of SER+, including a higher frequency of aberrant spindle formation and an elevated incidence of cytokinesis failure in embryos derived from SER+ oocytes [31]. We demonstrated that SER+ had a positive effect on oocyte maturation rate but did not affect births. We hypothesize that this is because the best embryos are chosen at each step in the process, and the most poor-quality oocytes (including those designated as showing SER+) are removed.

As we were not able to observe SER+ in oocytes under light microscopy, we turned to electron microscopy and noted aggregates of 2–9 μm [2]. The results suggested that SER+ is more common than previously noted due to its low visibility under light microscopy.

Oligonucleotide microarray analysis of SER+ oocytes versus SER− oocytes showed six down-regulated genes (CROCC, FDXR, HAUS8, MAP 2, MRPL11, and RPS3) and three up-regulated genes (GPSM1, GPSM3, and RAP1GAP) [32]. The gene-expression changes were determined to be involved in (i) cell and mitotic/meiotic nuclear division, (ii) organization of cytoskeleton and microtubules, and (iii) mitochondrial structure and activity [32]. None of these changes were linked to ER stress.

Limitations

For this study, data were collected between January 1, 1978, and August 31, 2020—with a SER cycle having one or several SER oocytes and the remaining oocyte(s) classified as normal [4]. Since SER+ embryos are not the typically preferred embryos for transplantation, many of them are not transferred [3]. In addition, when SER+ MII oocytes are used in a grouping experiment, they are evaluated again before ET [21]; thus, the significance of this study may be limited.

A study in which the outcomes of SER+ cycles/oocytes were assessed indicated that fetal malformations with SER+ cycles were greater than for SER− cycles [6]. Therefore, although the SER+ cycle can be recovered after embryo transfer, the fetus may show congenital defects, and this should be considered when transplanting SER+ cycle embryos.

Conclusions

We agree with the recent Alpha/ESHRE consensus [1] that transplantation of embryos with SER+ should be carefully considered. Embryos produced from SER+ cycles can be used but should only be transferred if no other suitable embryos are available over several cycles. This may provide the only prospect for completing a pregnancy to term with the birth of a healthy baby, and we feel this is a choice patients should have the right to make. The present technology requires patient consent because the patient must be made aware of the risk of complex chromosomal rearrangements with consistent 2q31 deletions.