The impact of the origin of surgical sperm retrieval on placental and embryonic development: The Rotterdam Periconception cohort

Abstract Background In patients with azoospermia, pregnancy can be achieved after surgical techniques using sperm retrieved from the testis or epididymis, which can impact on DNA integrity and epigenetics. DNA of the fetus and placenta is equally derived from both parents; however, genes important for placental development are expressed from the paternal alleles. Therefore, the origin of sperm may affect fetal and placental development. Objectives To investigate whether first‐trimester trajectories of embryonic and placental development of pregnancies conceived after intracytoplasmic sperm injection (ICSI) with testicular sperm extraction (TESE) or microsurgical epididymal sperm aspiration (MESA), are different from pregnancies after ICSI with ejaculated sperm or natural conceptions. Materials and methods A total of 147 singleton ICSI pregnancies, including pregnancies conceived after TESE (n = 23), MESA (n = 25) and ejaculated sperm (n = 99), and 380 naturally conceived and 140 after IVF treatment without ICSI were selected from the prospective Rotterdam periconception cohort. Crown‐rump length (CRL), embryonic volume (EV), Carnegie stages, and placental volume (PV) at 7, 9, and 11 weeks of gestation were measured using 3D ultrasound and virtual reality technology. Results Linear mixed model analysis showed no differences in trajectories of CRL, EV, and Carnegie stages between pregnancies conceived after ICSI with testicular, epididymal, and ejaculated sperm. A significantly positive association was demonstrated for PV between pregnancies conceived after TESE‐ICSI (adjusted beta: 0.28(95%CI: 0.05‐0.50)) versus ICSI with ejaculated sperm. Retransformation to original values showed that the PV of pregnancies after TESE‐ICSI is 14.6% (95%CI: 1.4%‐25.5%) larger at 11 weeks of gestation compared to ICSI pregnancies conceived with ejaculated sperm. Discussion and Conclusion Here we demonstrate that the first‐trimester growth trajectory of the placenta is increased in pregnancies conceived after TESE‐ICSI compared to those conceived after ICSI with ejaculated sperm. Findings are discussed in the light of known differences in sperm DNA integrity, epigenetics, and placental gene expression.


| INTRODUC TI ON
Male factor subfertility is an increasing problem due to aging, obesity, and poor lifestyle, which can often be treated with lifestyle interventions. 1,2 In severe male factor subfertility, however, intracytoplasmic sperm injection (ICSI) is often successfully used to achieve pregnancy. Sperm used for assisted reproductive treatment (ART) can have different origins dependent on the underlying cause or diagnosis of the male factor. In cases of non-obstructive azoospermia (NOA) sperm can be retrieved surgically by testicular sperm extraction (TESE). TESE can also be performed in the case of post-vasectomy, iatrogenic, congenital or post-infectious obstructive azoospermia (OA), or when vasovasostomy or microsurgical epididymal sperm aspiration (MESA) failed. 3 In both cases the sperm is not (TESE) or only partially (MESA) transported through the epididymis. It is becoming increasingly clear that this transport through the epididymis is a key factor in sperm maturation and functioning. 4 The epididymis secretes epididymal specific proteins, hormones, small non-coding RNAs and these factors combined can influence sperm epigenetics, gene expression and modify the sperm surface. 5 Although testicular sperm has not entered the anatomical part of the epididymis, it has the ability to activate the oocyte after ICSI. However, testicular sperm morphological quality is usually low and several studies indicate that the incidence of chromosomal abnormalities is increased in testicular sperm. 6,7 These abnormalities are associated with a decreased chance of implantation and lower ongoing pregnancy rates, possibly due to an increased aneuploidy rate in embryos. 7,8 However, a recent study in more than 340.000 IVF/ICSI cycles showed no clinical differences between pregnancies conceived with testicular, epididymal or ejaculated sperm regarding pregnancy rate and full-term delivery. 9 To overcome issues of timing and multiple surgical procedures for men, sperm can also be cryopreserved and thawed for later use.
Several studies show similar treatment outcomes between fresh and frozen sperm. 10,11 The period of embryonic and placental growth in the first trimester of pregnancy is characterized by rapid cell multiplication and therefore vulnerable for alterations in both maternal-and paternal-originated DNA integrity and epigenetic (re)programming.
Interestingly, paternally imprinted genes are predominantly expressed in the placenta. 12 Embryonic and placental growth can be reliably measured using three-dimensional ultrasound (3D-US) techniques in combination with virtual reality technology. 13 This combination allows in-depth perception and assessment of growth by measuring crown-rump length (CRL) and embryonic volume (EV). Furthermore, it is possible to make an accurate assessment of the Carnegie stage based on internal and external morphological characteristics, as a marker of embryonic development. 14 Since there are a number of differences regarding chromosomal constitution and epigenetics between sperm retrieved after TESE and MESA as compared to ejaculated sperm combined with a large preferential paternal expression profile of the placental genome, we hypothesize that embryonic and placental growth and development might differ between these groups.  15 Women and their partners were eligible for inclusion if they were at least 18 years of age and had an ongoing intrauterine singleton pregnancy and were less than 10 + 0 weeks of gestational age (GA). Participants were recruited for inclusion from November 2010 onwards.
For the current analysis, we included pregnancies of women who conceived after ICSI in combination with testicular, epididymal, cryopreserved or ejaculated sperm. For a general reference group, we included pregnancies of women who conceived through the light of known differences in sperm DNA integrity, epigenetics, and placental gene expression.

K E Y W O R D S
fertilization in vitro, placenta, spermatozoa, testicular sperm, trophoblast IVF with ejaculated sperm and women who conceived naturally.
GA was either based on the exact conception date for ICSI and the reported last menstrual period for naturally conceived pregnancies. We excluded pregnancies conceived after using donor semen or oocyte donation, and pregnancies complicated by congenital malformations and intrauterine fetal demise. Furthermore, we excluded pregnancies of women with an irregular menstrual cycle (menstrual cycle of less than 25 days or more than 31 days) or from which no first day of the last menstrual period was known, since in these pregnancies GA is based on CRL, which is our main outcome parameter.
In general, an obstructive component was excluded on medical history, physical examination, reproductive hormones (luteinizing hormone (LH), follicle-stimulating hormone (FSH), and testosterone) and scrotal ultrasound. Karyotyping and Y-chromosomal microdeletions were determined in all of these patients. All men with a diagnosed obstructive azoospermia (OA) initially underwent MESA. When motile sperm were microscopically detected, sperm cryopreservation was performed. When no motile sperm was found, TESE was performed. The latter group was excluded for analysis to create a homogenous TESE-ICSI group with only NOA patients.

| Study parameters
Baseline characteristics and obstetric history were retrieved through self-administered questionnaires covering details on age, ethnicity, and educational level. All data were verified at study entry by a researcher. Anthropometrics were measured by a researcher at study entry. Smoking and alcohol were defined as any consumption during the periconception period. Details regarding subfertility diagnoses, method of sperm retrieval and whether IVF or ICSI was used, were retrieved from the electronic patient files.

| Ultrasound data
From November 2010 onwards women underwent three transvaginal ultrasounds, in the 7th, 9th, and 11th week of gestation.
Ultrasound scans were performed with a 6-12 MHz transvaginal probe using GE Voluson E8 equipment and 4D View software (General Electrics Medical Systems, Zipf, Austria).
To optimally make use of the depth information present in 3D-ultrasound data, images were transferred to our Barco I-Space (a Cave Automatic Virtual Environment-like virtual reality system). 13 In the Barco I-space, an interactive virtual reality hologram, which allows depth perception, was created. All measurements were performed by trained research staff.

| Outcome variables
Outcome variables were generated multiple times in the first trimester of pregnancy, which gives the opportunity to study growth trajectories over time. CRL was measured three separate times per time point and embryo, and the mean of these measurements was used for analysis. EV measurements were performed once at each time point per embryo using a semiautomatic method based on gray levels. 16 Carnegie stages were determined once per time point to assess external morphological features of the embryo such as the development of the limbs and the curvature of the embryo, according to the protocol which was published previously. 14 Placental volume (PV) in first trimester, also known as trophoblast volume, was measured once per time point per pregnancy using Virtual Organ Computer-aided AnaLysis (VOCAL) (TM; GE Medical Systems, Zipf, Austria). 17 In short, twelve sections of the placenta were obtained using a rotational step of 15º. Trophoblast and myometrium can be distinguished by their difference in echogenicity, thereby calculating the total pregnancy volume. The volume of the placenta can be calculated by subtracting the gestational sac volume from the total pregnancy volume.

| Statistical analysis
To take the correlation between measurements of the same pregnancy into account, we used a linear mixed model. In the first trimester of pregnancy, we used this linear mixed model to assess the associations between the different origins of sperm and CRL, EV, Carnegie stage, and PV. Furthermore, we used naturally conceived pregnancies and pregnancies after IVF with ejaculated sperm as a reference group for the different origins of sperm for ICSI. In the first model, we adjusted for GA only. In the second model, we additionally adjusted for the paternal covariates age, smoking and alcohol use and maternal parity selected based on the characteristics of the study groups and literature.
p-values <0.05 were considered statistically significant. All analyses were performed using SPSS package 24.0 (IBM SPSS Statistics, Armonk, NY).

| Baseline
From a total of 1,743 pregnant women included in the cohort, 1,289 participated together with their male partner. Due to missing ultrasound data and women who did not undergo a first-trimester ultrasound, 213 pregnancies were excluded. Furthermore, we excluded pregnancies because of miscarriage (n = 87), oocyte donation (n = 5), congenital malformations (n = 18), intrauterine fetal death (n = 10), and termination of pregnancy (n = 11) and cases with an irregular menstrual cycle prior to natural conception (n = 257). The remaining 688 included pregnancies comprised of 380 naturally conceived pregnancies, 140 pregnancies after IVF, and 168 pregnancies after ICSI.
Of the included ICSI pregnancies, 28 were conceived after ICSI with At baseline, there were no significant differences regarding paternal age, BMI, geographical background, education, alcohol use, and smoking between the different ICSI groups. Paternal age was significantly lower in the naturally conceived pregnancy group compared to the IVF group (33.6 vs 36.2 year respectively (P < .01) ( Table 1). Regarding maternal factors at baseline, we found no statistical differences for age, BMI, geographical origin, education, and alcohol use. There was a significant difference regarding parity, where the percentage nulliparous in the naturally conceived group was lower compared to all other groups (P < .001).
Of the included pregnancies resulting after TESE-procedures, the indication for ICSI treatment was because of either male (82.6%) or combined male-female factor subfertility (17.4%), with 100% being a non-obstructive azoospermia (Table S1).
No differences were seen regarding the embryonic growth parameters CRL and EV when comparing the TESE and MESA-ICSI groups with freshly ejaculated sperm-ICSI (Table 2 and Figure 2B).  (Table 2). After adjustment for paternal covariates age, smoking, alcohol, and maternal parity (Model 2) the significantly positive association remains (beta: 0.28 (95%CI: 0.05 to 0.50)).

Retransformation of the betas to the original values showed that
PVs of pregnancies after TESE-ICSI are 14.6% (95%CI: 1.4% to 25.5%) larger at 11 weeks of gestation as compared to pregnancies after fresh ejaculated-ICSI (Figure 2A).
To investigate the effect of cryopreservation, we pooled the groups with TESE and MESA treatment and cryopreserved sperm, and compared them with the group of ICSI with freshly ejaculated sperm. Again, we found no statistically significant differences regarding trajectories of CRL, EV, PV, and Carnegie stage (Table 3 b).

| DISCUSS ION
In this study, we show that pregnancies conceived after TESE-ICSI compared with ejaculated sperm-ICSI are associated with trajectories of increased PV in the first trimester of pregnancy, with an   All testicular and epididydimal extracted sperm is cryopreserved after collection and thawed at the moment of oocyte pickup.
Cryopreservation of sperm has been found to significantly alter sperm DNA methylation. 23,24 However, despite these reported epigenetic variations, we found no differences regarding embryonic and placental growth and development after pooling of all groups using cryopreserved sperm (TESE, MESA, and cryopreserved sperm) compared to ICSI with freshly ejaculated sperm. We cannot distinguish between the influence of the origin of the sperm or cryopreservation since in our clinic cryopreservation of the sperm is always used after TESE and MESA.
The finding of larger PV trajectories in early pregnancies conceived after TESE-ICSI can also be explained by differences in DNA TA B L E 2 Betas of embryonic development (CRL, EV, Carnegie stages) and placental growth trajectories (PV) of different origins of sperm retrieval compared to ICSI with ejaculated sperm   integrity due to DNA damage. Although the epidydimis is thought to play an important role in sperm maturation, a recent meta-analysis showed that sperm DNA damage, measured by sperm DNA fragmentation, can be significantly higher in freshly ejaculated sperm compared to testicular sperm. 25 The lower sperm DNA fragmentation can explain the larger placental development, in terms of better development with improved sperm DNA integrity. PV is associated with birth outcomes: pregnancies ending in miscarriage had smaller placental volumes during the first trimester as compared with those that result in a livebirth, indicating the potential beneficial role of using testicular sperm, since we found placental volume to be increased. 17,26 A confounding factor can be maternal smoking, since significantly more women smoked in the TESE-ICSI group. However, since smoking is associated with smaller placental growth in the first trimester of pregnancy, PV is hypothesized to be even larger after using testicular sperm for ICSI in women who do not smoke.
Previous research investigating the association between surgically retrieved sperm and pregnancy outcomes comprised of far more participants than our study and mentioned no significant differences between testicular, epididymal, and ejaculated sperm regarding the birth outcomes birthweight and preterm birth. [27][28][29] The aim of our study was to gain more insight into the early (patho) physiology of the role of the origin of sperm and embryonic and placental development, not to study pregnancy outcomes. Our study was not powered to detect differences regarding pregnancy outcomes. The posthoc sample size calculation using an α-level of 0.05 and power of 80%, revealed that at least 80 participants in the total study group are needed to accurately show significance regarding placental volume, which were present in our study. This study revealed the magnitude of effect sizes which will aid to determine an optimal sample size for future larger studies or randomized controlled trials.
No effects were established regarding embryonic growth, as measured by CRL and EV, between the different sperm origins. Since paternally expressed genes are predominantly expressed in the placenta, we expected to show periconceptional paternal effects on PV trajectories. Our group previously showed that IVF-ICSI pregnancies exhibit larger embryonic growth compared to naturally conceived pregnancies indicating that the procedure itself can induce differences regarding embryonic development. 30 Another possible explanation for a lack of detectable effect of TESE sperm on CRL and EV, could be due to the strong impact of the IVF-ICSI procedure itself on epigenetic reprogramming of the embryo and endometrial receptivity. 31,32 Our study has several strengths and limitations. A strength is the availability of multiple serial 3-D ultrasounds in the first trimester of the same pregnancy and the possibility to precisely assess several morphogenic features of embryonic and placental growth and development. Our group previously showed that early PV can reliably be measured with very high intra-class correlation coefficients (ICC > 0.95). 33 In the present study, we confirm other studies that showed no significant differences regarding PV between naturally conceived pregnancies (n = 84) and IVF-ICSI pregnancies (n = 70) (39.8cm 3 and 40.2cm 3 respectively). 34   OA in the MESA group), other differences between groups are present, such as the cause of infertility. Since our study population is too small to correct or stratify for these factors, future larger studies should incorporate these factors in sample size calculation and analyses. Furthermore, we did not investigate birth outcomes, which was not an outcome in this study and therefore not powered accordingly.

| CON CLUS ION
We show that placental growth trajectories in the first trimester of pregnancy are increased in pregnancies conceived after TESE-ICSI as compared to ejaculated sperm. No significant differences are shown regarding embryonic development measured as CRL, EV and Carnegie stages. These findings might be partially explained by differences in DNA damage, chromosomal constitution and epigenetics of testicular sperm as compared to ejaculated sperm. After this explorative study, future research should validate these findings in larger cohorts, including investigating possible associations with birth outcomes.
Exploring the underlying pathofysiological mechanisms by measuring sperm DNA damage, placental and also the neonatal epigenome will provide insights that help optimize preconceptional health and counseling. This will help to further improve pregnancy chances and birth outcomes in subfertile couples with male factor subfertility.

ACK N OWLED G EM ENTS
The Rotterdam periconception cohort (Predict Study) is conducted by the department of Obstetrics and Gynecology at the Erasmus MC, University Medical Center, Rotterdam. We gratefully acknowledge the Predict team for data acquisition and the participating couples, gynecologists, and midwifery practices in Rotterdam for their contributions.

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest.

AUTH O R ' S CO NTR I B UTI O N S
RST and SS initiated the research question and supervised all aspects of the study. EB was responsible for the ICSI treatment in the