The effect of ICSI in infertility couples with non-male factor: a systematic review and meta-analysis

Abstract

Purpose

We performed a systematic review and meta-analysis of available literature to investigate the efficacy of the intracytoplasmic sperm injection (ICSI) in couples with non-male factor with respect to the clinical outcomes.

Methods

The literature search was based on EMBASE, PubMed, and the Cochrane Library. All studies published after 1992 until February 2020 and written in English addressing patients in the presence of normal semen parameters subjected to ICSI and in vitro fertilization (IVF) were eligible. Reference lists of retrieved articles were hand-searched for additional studies. The primary outcomes were fertilization rate, clinical pregnancy rate, and implantation rate; the secondary outcomes were good-quality embryo rate, miscarriage rate, and live birth rate.

Results

Four RCTs and twenty-two cohort studies fulfilling the inclusion criteria were included. Collectively, a meta-analysis of the outcomes in RCTs showed that compared to IVF, ICSI has no obvious advantage in fertilization rate (RR = 1.16, 95% CI: 0.83–1.62), clinical pregnancy rate (RR = 1.04, 95% CI: 0.66–1.64), implantation rate (RR = 1.12, 95% CI: 0.67–1.86), and live birth rate (RR = 1.17, 95% CI: 0.43–3.15). Pooled results of cohort studies demonstrated a statistically significant higher fertilization rate (RR = 1.16, 95% CI: 1.03–1.31) and miscarriage rate (RR = 1.04, 95% CI: 1.01–1.06) in the ICSI group; furthermore, higher clinical pregnancy rate (RR = 0.85, 95% CI: 0.77–0.94), implantation rate (RR = 0.78, 95% CI: 0.65–0.95), and live birth rate (RR = 0.86, 95% CI: 0.79–0.94) was founded in the IVF group; no statistically significant difference was observed in good-quality embryo rate (RR = 0.98, 95% CI: 0.93–1.04).

Conclusion

ICSI has no obvious advantage in patients with normal semen parameters. Enough information is still not available to prove the efficacy of ICSI in couples with non-male factor infertility comparing to IVF.

Electronic supplementary material

The online version of this article (10.1007/s10815-020-01970-9) contains supplementary material, which is available to authorized users.

Introduction

Intracytoplasmic sperm injection (ICSI), a kind of assisted reproductive technology (ART), has become the most commonly used method of fertilization. It was primarily introduced to manage male infertility with severely impaired characteristics like nonobstructive azoospermia or no measurable sperm count [1, 2]. Although the reliability of ICSI has made it an attractive option even in non-male factor couples suffering infertility worldwide, as considered more invasive and costly than conventional in vitro fertilization (IVF), ICSI has been a controversial option for these couples with non-male factor indications [3, 4]. Furthermore, the arbitrary selection of the inseminating spermatozoon may give rise to other detrimental effects [5]. Pregnancies resulting from the use of ICSI have been found to be associated with increased incidences of congenital malformation [6, 7], chromosomal aberrations [8], and birth defects [9] compared to in vitro fertilization.

Despite the uncertainness, frequently, ICSI use has expanded to other indications, including infertility of unexplained cause, low oocyte yield, advanced female age, previously failed IVF cycles, and so forth [10–15]. It has become popularized to the stage of routine laboratory service and it is even implemented for infertility couples with normal semen parameters throughout the world. ICSI without male factor infertility may be of benefit for select patients undergoing IVF with preimplantation genetic testing and previously cryopreserved oocytes according to the Practice Committee of the American Society for Reproductive Medicine (ASRM) [16]. ICSI use remarkably increased in recent years and it should be noted that the relative increase was among patients without male factor infertility, without sufficient evidence of a benefit to use ICSI over in vitro fertilization (IVF) [17–19]. Multiple studies have been carried out on this subject recently, where reporters came to a controversial conclusion. Some studies did not show any benefit on clinical outcomes in these couples undergoing ICSI, whereas other studies demonstrated that clinical outcomes in ICSI were preferable between couples affected with non-male factors when compared with conventional IVF. It has been reported that ICSI not only improved reproductive outcomes but also reduced total fertilization failure among patents with non-male factor infertility [20, 21]. ICSI has been suggested in alternative to IVF for a woman with fertilization failure in previous IVF cycles, due to subsequent ICSI may help to improve fertility rate [22]. However, the conclusions are not coincident, and recent studies challenge the application of ICSI for non-male indications [23–25]. The application of ICSI to treat patients with non-male factor is still controversial and clinical evidence is currently lacking. Although ICSI may have a role in improving clinical outcomes, the routine use of ICSI for these indications requires further investigation.

Therefore, we conducted this systematic review and meta-analysis to review recent literature regarding clinical outcomes following ICSI and IVF in infertility couples in the presence of normal semen parameter and to evaluate the effectiveness of ICSI in these infertility couples. Our primary clinical outcomes were fertilization rate (FR), clinical pregnancy rate (CPR), and implantation rate (IR). And, the secondary outcomes were birth rate (LBR), miscarriage rate (MR), and good-quality embryo rate.

Materials and methods

Literature search and study selection

This systematic review and meta-analysis were carried out in accordance with the PRISMA guidelines and had been registered on the International Prospective Register of Systematic Reviews Platform (PROSPERO) (CRD42020157499). A comprehensive PubMed, Embase, and Cochrane search was performed to find studies aimed to compare clinical outcomes after ICSI and IVF of infertility couples in the presence of normal semen parameters and restricted to papers fully published in English dating from 1992 until February 2020 using the following keywords: intracytoplasmic sperm injection, in vitro fertilization, non-male factor, normozoospermia, normozoospermic semen. The full search strategy was provided in supplementary data. Reference lists of potential studies were also searched for additional information. We excluded the following studies: (1) male patient with abnormal semen parameters including borderline semen as defined by WHO criteria or local laboratory standards in each center; (2) studies including rescued ICSI; (3) unavailable raw data, overlapped data sets books; (4) review articles, case reports, and study without full text. The identification of relevant studies was conducted by three of the authors independently, and discussion was needed if there were any discrepancies.

Data extraction

Data were extracted from the specific papers and not available data were requested by contacting authors if necessary. All data were abstracted and put into a table format in a systematic manner. The following study characteristics were obtained: methods (study design, study duration, study unit, sample size), participants (female age, cause of infertility), and raw data of outcomes (fertilization rate, clinical pregnancy rate, implantation rate, live birth rate, good-quality embryo rate, and miscarriage rate).

Quality assessment

Newcastle–Ottawa Scale was employed to evaluate the quality of individual observational study as a qualitative assessment tool. The scoring system is based on three dimensions covering the selection of groups, the comparability of groups, and the outcome of interest [26]. Meanwhile, the risk of bias of RCTs was assessed according to the Cochrane Collaboration’s tool [27]. Quality assessment was carried out independently by three of the authors and disagreements were resolved through group discussion. When 10 or more studies were included in one analysis, the funnel plot would be constructed in order to assess possible publication bias with Egger’s test [28].

Statistical analysis

Comparative analysis was performed between the two groups and the effect of ICSI and IVF was assessed according to study designs separately and random-effect models were used to pool outcomes. For dichotomous data (for example, fertilization rate), the raw data in the two groups were used to calculate the Mantel–Haenszel relative risk (RR) and 95% CI. The I² statistical test was employed to assess heterogeneity. When I² ≥ 50%, it was considered that there was substantial heterogeneity with the included studies. Further subgroup analyses were performed to identify potential sources of heterogeneity when heterogeneity was considered to be substantial. Sensitivity analysis was undertaken for each outcome to determine whether the conclusions were robust. Data were analyzed using Review Manager 5.3 and R software (version 3.5.1).

Results

Results of the searches

A total of 1599 results yielded through database and other sources. After removing the duplicates, 962 articles were screened by title and abstract by three reviewers independently, and 881 articles that did not meet the eligibility criteria were excluded. And, 54 of the remaining studies were excluded according to different reasons. Thus, 26 studies were included in the meta-analysis. A flow diagram of the searching process of these records is shown in Fig. 1.

Fig. 1.

PRISMA flow diagram of study selection for the systematic review and meta-analysis

Characteristics of included studies

A total of 26 studies (4 RCTs and 22 cohort studies) met the aforementioned eligibility criteria and were included in this study. A total of 26,120 patients were included except three of studies did not mention sample size of patients [33, 36, 46]. The mean participant age was 29 to 41 years old across the studies. Study characteristics including study design, study unit, cause of infertility, female age, study duration, study unit, number of patients/cycles, and primary outcomes were summarized in Table 1.

Table 1.

Characteristics of studies included in the meta-analysis

Author, year, country	Study design	Duration	Patient characteristics	Study unit	Patients (n)	Cycles (n)	Female age (y)	Primary outcome
Aboulghar et al. [29] (1996), Egypt	Prospective cohort study	NA	Unexplained infertility	Oocyte	22	NA	33 ± 4.2	FR
Aboulghar et al. [30] (1996), Egypt	RCT	NA	Tubal	Patient	116	116	35.6 ± −5.0	FR, CPR, IR
Artini et al. [31] (2013), Italy	Retrospective cohort study	January 2007 to July 2012	Severe poor responder patients (n = 386)	Cycle	386	425	38.23 ± 3.82	FR, CPR, good-quality embryo rate, IR, MR
Biliangady et al. [32] (2019), India	Retrospective cohort study	January 2012 to December 2017	Woman aged 25 to 35 years with normal ovarian reserve	Patient	350	350	31.20 ± 3.76	CPR, IR, MR, LBR,
Boulet et al. [33] (2015), Georgia	Retrospective cohort study	2008 to 2012	Tubal factor, endometriosis, uterine factor, ovulatory disorder, diminished ovarian reserve, unexplained, and others	Cycle	NA	317,996	NA	CPR, IR, LBR
Bukulmez et al. [34] (2000), Turkey	RCT	NA	Tubal factor	Patient	76	76	34.02 ± 3.84	FR, CPR, IR
Eftekhar et al. [35] (2012), Iran	Retrospective cohort study	April 2009 to September 2010	Ovarian, tubal factor, mild endometriosis, unexplained, uterine, mixed	Patient	220	200	29.28 ± 3.86	CPR
Fang et al. [14] (2012), China	Retrospective cohort study	January 2007 to December 2010	Patients with poor ovarian reserve in whom only one or two oocytes were retrieved	Patient	194	194	36.34 ± 4.60	PR, CPR
Farhi et al. [15] (2019), Israel	Retrospective cohort study	NA	All aged ≥ 35 (unexplained, tubal factor, and patients with endometriosis)	Oocyte	52	52	38.7 ± 2.6	FR
Fernández et al. [36] (2014), Spain	Retrospective cohort study	January 2011 to April 2013	Women with low response (oocyte number at pick up < 6	Cycle	NA	590	33.2	IR, CPR, MR
Grimstad et al. [37] (2016), USA	Retrospective cohort study	2004–2012	Tubal ligation	Patient	7145	7145	35.57 ± 4.01	CPR,LBR
Hershlag et al. [38] (2002), USA	Prospective cohort study	January 1998 to March 30, 2001	Unexplained infertility	Oocyte	60	NA	NA	FR
Hwang et al. [39] (2005), Taiwan	RCT	January 1998 to January 2000	PCOS	Oocyte	60	60	32.5 ± 2.3	FR
Jaroudi et al. [40] (2003), Saudi Arabia	Retrospective cohort study	June 1997 and April 2002	Unexplained infertility	Oocyte and Patient	199	NA	NA	FR
Khamsi et al. [41] (2001), Canada	RCT	NA	Tubal (n = 18), unknown, endometriosis, oocyte donation, therapeutic donor insemination	Oocyte	35	NA	33.9 ± 5.46	FR, good-quality embryo rate
Kim et al. [42] (2014), Korea	Retrospective cohort study	April 2009 to March 2013	Tubal factor (n = 94), age factor (n = 57), endometriosis (n = 47), polycystic ovary syndrome (n = 20), and unexplained infertility (n = 78)	Cycle	217	296	35.53 ± 4.51	FR
			Patients with three or less oocytes retrieved	Cycle	88	100	NA	CPR, IR, MR, good-quality embryo rate
Komsky-Elbaz et al. [21] (2013), Israel	Retrospective cohort study	January 2000 to December 2011	Endometriosis	Oocyte	35	79	30.9 ± 3.5	FR
Ming et al. [43] (2015), China	Retrospective cohort study	January 2011 and December 2013	Tubal factor, endometriosis, ovulatory dysfunction	Oocytes	218	218	32.9 ± 4.1	FR, good-quality embryo rate
Li et al. [24] (2018), Australia	Retrospective cohort study	July 2009 to June 2014	Female factor infertility and unexplained infertility	Patient	14,693	14,693	NA	LBR
Liu et al. [44] (2018), China	Retrospective cohort study	June 2011 to May 2016	Women with age ≥ 40 years and ≤ 5 oocytes obtained	patient	644	644	41.34 ± 1.07	FR, IR, CPR LBR, good-quality embryo rate
Van der Westerlaken et al. [45] (2005), The Netherlands	Prospective cohort study	September 1995 to February 2003	First IVF attempt with total fertilization failure or with low fertilization (< 25%)	oocyte	38	38	32.2 ± 3.9	FR, CPR
Schwarze et al. [46] (2017), Chile	Retrospective cohort study	January 2012 to December 2014.	With non-male factor	Cycle	NA	49,813	36.78 ± 4.41	MR, LBR
Shveiky et al. [13] (2006), Israel	Retrospective cohort study	January 1999 to December 2002	Unexplained infertility	Cycle	118	118	31.5 ± 5.6	FR, IR, PCR, LBR
Tannus et al. [47] (2017), Canada	Retrospective cohort study	January 2012 to June 2015	Women aged 40–43 years	Patient	745	745	41.17 ± 0.90	CPR, LBR
Xi et al. [48] (2012), China	Retrospective cohort study	January 2009 to December 2010.	Patients in which no more than three oocytes	Cycle	396	406	NA	FR, CPR, IR, MR
Yang et al. [49] (1996), Qatar	Retrospective cohort study	June 1994 to September 1994	Tubal = 10 unexplained infertility	Oocyte	13	13	29	FR

In this meta-analysis, quality assessment for RCTs was performed according to the Cochrane Collaboration’s tool and was summarized in Fig. 2. Both study design and methodology quality varied among the included four studies. Only one RCT used the optimal method of sequence generation and described exhaustively, whereas the other three studies generate the sequence in an inappropriate way or did not describe sequence generation explicitly. All the studies did not describe the method of blinding. For all cohort studies, the Newcastle–Ottawa Scale was employed to assess the quality and the NOS scores ranged from 5 to 9; the results were summarized in Table 2. All the studies provided information on ICSI and IVF groups, and most were similar. Three of them had a medium risk of bias and remained 19 studies had low risk.

Fig. 2.

Diagram of the quality of included RCTs. Each risk of bias item presented as percentages across all included studies (a) and methodological quality items (b)

Table 2.

Newcastle–Ottawa Quality Assessment Scale of included observational studies

Study	Design	Selection				Comparability	Outcome			Summary score (risk of bias)	Final quality assessment
		Representativeness of the exposed cohort	Selection of the non-exposed cohort	Ascertainment of exposure	Demonstration that the outcome of interest was not present at the start of the study	Comparability of cohorts on the basis of the design or analysis	Assessment of outcome	Was follow-up long enough for outcomes to occur	Adequacy of follow-up of cohorts
Aboulghar et al. [29]	Prospective cohort study	a*	a*	a*	a*	a*	b*	b	a*	7/9 (low)	High quality
Artini et al. [31]	Retrospective cohort study	b*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Biliangady et al. [32]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Boulet [33]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Eftekhar [35]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Fang et al. [14]	Retrospective cohort study	b*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Farhi et al. [15]	Retrospective cohort study	b*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Fernández et al. [36]	Retrospective cohort study	a*	a*	d	b	a*	d	a*	a*	5/9 (medium)	Intermediate quality
Grimstad et al. [37]	Retrospective cohort study	a*	a*	d	b	a*	d	a*	a*	5/9 (medium)	Intermediate quality
Hershlag [38]	Prospective cohort study	a*	a*	a*	a*	a*	b*	a*	a*	8/9 (low)	High quality
Jaroudi \[40]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Kim et al. [42]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Komsky-Elbaz et al. [21]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Ming [43]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Li et al. [24]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Liu et al. [44]	Retrospective cohort study	b*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Van der Westerlaken et al. [45]	Prospective cohort study	b*	a*	a*	a*	a*	a*	a*	a*	8/9 (low)	High quality
Schwarze et al. [46]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Shveiky et al. [13]	Retrospective cohort study	a*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Tannus et al. [47]	Retrospective cohort study	b*	a*	a*	b	a*	b*	a*	a*	7/9 (low)	High quality
Xi et al. [48]	Retrospective cohort study	b*	a*	a*	b	a*	d	a*	a*	6/9 (medium)	Intermediate quality

* meaned one star according to the NOS useing a ‘star‐based’ rating‐system and the studies scored from 0 to 9

Meta-analysis of fertilization rate

A total of 16 studies (4 RCTs and 12 cohort studies) evaluated the outcome of fertilization rate between the ICSI group and the IVF group. Subgroup analysis was performed based on study design (Fig. 3). The pooled outcome from 4 RCTs showed an increased fertilization rate in the ICSI group, but there was no significant difference between the two groups (RR = 1.16, 95% CI: 0.83–1.62). Significant heterogeneity was observed within studies (I² = 97%, P < 0.01). However, when the outcome was pooled for the cohort studies, there was a significant higher fertilization rate observed in the ICSI group (RR = 1.16, 95% CI: 1.03–1.31). But, significant heterogeneity was observed within studies (I² = 94%, P < 0.01).

Fig. 3.

Forest plot of studies of ICSI group versus IVF group for the outcome of FR

Meta-analysis of clinical pregnancy rate

Two RCTs and 13 cohort studies evaluated the outcome of the clinical pregnancy rate. Pooled analysis from cohort studies showed that IVF had an advantage in the clinical pregnancy rate than ICSI (RR = 0.85, 95% CI: 0.77–0.94) (Fig. 4). Furthermore, substantial heterogeneity was observed in cohort studies (I² = 66%, P < 0.01) and further subgroup analysis showed that CPR was higher in the IVF group among female age 35 and older (RR = 0.73, 95% CI: 0.59–0.90), while there was no difference of CPR in female aged under 35 between two groups (RR = 0.87, 95% CI: 0.71–1.08). Women with age data not available were included in the third subgroup. Despite there was no difference observed in CPR (RR = 0.91, 95% CI: 0.43–1.90), high heterogeneity existed in these data (I² = 75%, P = 0.02) (Fig. 5). However, when the outcome was pooled for 2 RCTs, no significant difference was observed (RR = 1.04, 95% CI: 0.66–1.64) (Fig. 4).

Fig. 4.

Forest plot of studies of ICSI group versus IVF group for the outcome of CPR

Fig. 5.

Subgroup analysis of cohort studies of ICSI group versus IVF group for the outcome of CPR according to age

Meta-analysis of implantation rate

Eight cohort studies and 1 RCT provided implantation rate data and were included in this meta-analysis [13, 29, 31–33, 36, 42, 44, 48]. The analysis of the single RCT revealed that there was no difference in implantation rate between the IVF group and the ICSI group (RR = 1.12, 95% CI: 0.67–1.86). The pooled results of cohort studies suggested implantation rate was significantly higher in the IVF group. But, moderate heterogeneity was found among the cohort studies (I² = 58%, P = 0.02) (Fig. 6). Subgroup analysis according to female age demonstrated that IR was significantly higher in the IVF group among female aged 35 and older (RR = 0.52, 95% CI: 0.36–0.74); nevertheless, the advantage was not observed in another subgroup where women aged under 35 (RR = 0.93, 95% CI: 0.77–1.12). No heterogeneity was found in both these subgroups (I² = 0%, P = 0.89; I² = 0%, P = 0.46). In addition, women with age data not available were included in the third subgroup; high heterogeneity was observed in these data set (I² = 82%, P = 0.02) (Fig. 7).

Fig. 6.

Forest plot of studies of ICSI group versus IVF group for the outcome of IR

Fig. 7.

Subgroup analysis of cohort studies of ICSI group versus IVF group for the outcome of IR according to age

Meta-analysis of live birth rate

Nine studies included in the analysis provided live birth rate data (1 RCT and 8 cohort studies) [13, 24, 32–34, 37, 44, 47, 46]. There was only one RCT showed LBR data and no significant difference was observed between the groups (RR = 1.17, 95% CI: 0.43–3.15). However, pooled results from 8 cohort studies demonstrated LBR was significantly higher in the IVF group (RR = 0.86, 95% CI: 0.79–0.94) (Fig. 8). Meanwhile, high heterogeneity was noted among these studies (I² = 91%, P < 0.01). Further subgroup analysis according to female age (< 35, ≥ 35) demonstrated that not only was the LBR significantly higher in the IVF group (RR = 0.93, 95% CI: 0.92–0.94; RR = 0.86, 95% CI: 0.78–0.83) but also the heterogeneities were not detected in both groups (I² = 0%, P = 0.54; I² = 0%, P = 0.49) (Fig. 9).

Fig. 8.

Forest plot of studies of ICSI group versus IVF group for the outcome of LBR

Fig. 9.

Subgroup analysis of cohort studies of ICSI group versus IVF group for the outcome of LBR according to age

Meta-analysis of miscarriage rate

Seven cohort studies provided miscarriage rate data and were included in the analysis [31–33, 36, 42, 44, 48]. No heterogeneity was noted among these studies (I² = 0%, P = 0.48). Pooled results revealed the miscarriage rate was higher in the ICSI group than the IVF group (pooled RR = 1.04, 95% CI: 1.01–1.06) (Fig. 10).

Fig. 10.

Forest plot of studies of ICSI group versus IVF group for the outcome of MR

Meta-analysis of good-quality embryo rate

Four cohort studies reported the good-quality embryo rate [14, 31, 42, 43]. The overall effect estimate demonstrated that ICSI did not significantly increase the rate of the good-quality embryo than IVF (RR = 0.98, 95% CI: 0.93–1.04) (Fig. 11). Low heterogeneity significant was observed (I² = 13%, P = 0.33).

Fig. 11.

Forest plot of studies of ICSI group versus IVF group for the outcome of good-quality embryo rate

Sensitivity analysis

Sensitivity analyses were undertaken with each study removed in turn for outcomes to determine whether the conclusions were robust to the arbitrary decision made regarding the eligibility and analysis of the studies (Tables 2 and 3). The outcomes of CPR, IR, LBR, and good-quality embryo rate did not vary markedly with the removal of any one study, demonstrating that the meta-analysis was robust, and the results were not overly influenced by any study. However, the outcome of FR and the miscarriage rate was not robust. A total of four included RCTs reported fertilization rate, after removing one of them [29]; the fertilization rate was statistically higher in the ICSI group (RR = 1.31, 95% CI: 1.06–1.62, P = 0.011), which was similar to the pooled results of cohort studies. In cohort studies, the result was not robust neither. Contrary to previous pooled results from cohort studies, removing the study conducted in 2005 [45] can lead to a different result, which demonstrated that there was no difference in FR between the two groups (RR = 1.10, 95% CI: 0.98–1.22, P = 0.097). There were seven cohort studies included to draw the result of the miscarriage rate. But, five of them can influence the final results [31, 33, 42, 44, 48] and give rise to no statistical difference (RR = 1.04, 95% CI: 0.91–1.21, P = 0.549; RR = 1.04, 95% CI: 0.91–1.19, P = 0.574; RR = 1.03, 95% CI: 0.93–1.15, P = 0.548; RR = 1.04, 95% CI: 0.89–1.22, P = 0.612).

Table 3.

Sensitivity analyses

Outcome		Study	RR	95% CI		P value
				Lower limit	Upper limit
Fertilization rate	RCTs	Aboulghar et al. [29]	1.31	1.06	1.62	0.011
		Hwang et al. [39]	1.04	0.79	1.38	0.767
		Khamsi et al. [41]	1.13	0.73	1.76	0.572
		Yang et al. [49]	1.17	0.76	1.80	0.469
	Cohorts	Aboulghar et al. [30]	1.18	1.04	1.34	0.011
		Artini et al. [31]	1.20	1.05	1.37	0.008
		Fang et al. [14]	1.15	1.02	1.30	0.028
		Farhi et al. [15]	1.14	1.01	1.29	0.034
		Hershlag [38]	1.14	1.01	1.29	0.030
		Jaroudi et al. [33]	1.17	1.02	1.33	0.022
		Kim et al. [42]	1.19	1.03	1.36	0.016
		Ming et al. [43]	1.20	1.07	1.33	0.001
		Liu et al. [44]	1.18	1.03	1.35	0.017
		Shveiky et al. [13]	1.19	1.04	1.35	0.012
		Van der Westerlaken et al. [45]	1.10	0.98	1.22	0.097
		Xi et al. [48]	1.16	1.02	1.32	0.022
Clinical pregnancy rate	RCTs	Aboulghar et al. [29]	1.00	0.42	2.39	1.000
		Bukulmez et al. [34]	1.06	0.62	1.80	0.842
	Cohorts	Artini et al. [31]	0.87	0.80	0.95	0.002
		Biliangady et al. [32]	0.85	0.77	0.94	0.002
		Boulet et al. [33]	0.79	0.68	0.92	0.003
		Eftekhar 2012	0.86	0.79	0.95	0.002
		Fang et al. [14]	0.85	0.77	0.94	0.001
		Fernández et al. [36]	0.84	0.76	0.93	< 0.001
		Grimstad et al. [37]	0.80	0.68	0.94	0.008
		Jaroudi [40]	0.85	0.77	0.93	< 0.001
		Kim et al. [42]	0.86	0.78	0.94	0.002
		Liu et al. [44]	0.86	0.78	0.95	0.002
		Shveiky et al. [13]	0.84	0.76	0.93	< 0.001
		Tannus et al. [47]	0.86	0.78	0.94	0.002
		Xi et al. [48]	0.86	0.79	0.95	0.002
Implantation rate	Cohorts	Artini et al. [31]	0.83	0.70	0.99	0.042
		Biliangady et al. [32]	0.75	0.59	0.96	0.021
		Boulet 2015 [33]	0.70	0.53	0.93	0.014
		Fernández et al. [36]	0.71	0.56	0.91	0.006
		Kim et al. [42]	0.79	0.65	0.97	0.023
		Liu et al. [44]	0.82	0.68	0.99	0.037
		Shveiky et al. [13]	0.75	0.60	0.94	0.011
		Xi et al. [48]	0.84	0.71	0.99	0.033
Live birth rate	Cohorts	Biliangady et al. [32]	0.87	0.79	0.95	0.003
		Boulet [33]	0.84	0.78	0.90	< 0.001
		Grimstad et al. [37]	0.87	0.78	0.97	0.009
		Li et al. [24]	0.85	0.76	0.94	0.002
		Liu et al. [44]	0.87	0.79	0.95	0.002
		Schwarze et al. [46]	0.89	0.83	0.95	0.001
		Shveiky et al. [13]	0.85	0.78	0.94	< 0.001
		Tannus et al. [47]	0.86	0.79	0.95	0.002
Miscarriage rate	Cohorts	Artini et al. [31]	1.04	0.91	1.21	0.549
		Biliangady et al. [32]	1.04	1.01	1.06	0.005
		Boulet 2015 [33]	1.07	0.74	1.54	0.737
		Fernández et al. [36]	1.04	1.01	1.07	0.003
		Kim et al. [42]	1.04	0.91	1.19	0.574
		Liu et al. [44]	1.03	0.93	1.15	0.548
		Xi et al. [48]	1.04	0.89	1.22	0.612
Good-quality embryo rate	Cohorts	Artini et al. [31]	1.01	0.96	1.06	0.837
		Fang et al. [14]	1.00	0.95	1.05	0.959
		Kim et al. [42]	0.97	0.90	1.05	0.478
		Ming et al. 2015 [43]	0.93	0.85	1.01	0.099

Assessment of publication bias

When 10 or more studies were included in one analysis, the funnel plot was constructed to examine possible publication bias with Egger’s test [28]. So, we only assessed the publication bias of outcomes regarding CPR and FR in this study. The funnel plot was symmetric for the outcome of CPR and Egger’s test did not suggest statistical significance (P = 0.0581) (Fig. 12), but the funnel plot was asymmetry for the outcome of FR, and Egger’s test showed statistical significance (P = 0.0018) (Fig. 13), indicating there was publication bias of included studies.

Fig. 12.

Funnel plot for CPR

Fig. 13.

Funnel plot for FR

Discussion

Primary outcomes

The aim of this systematic review and meta-analysis was to investigate whether ICSI improves clinical outcomes in couples with non-male factor indications. Our inclusion criteria were infertility couples in the presence of normal semen parameters, which distinguishes this study from the previous meta-analysis that included patients with broad line semen as well as mild factor infertility. The results of our study did not show ICSI yields a better clinical outcome than IVF in couples with non-male factor.

Due to clinical or methodological heterogeneity, RCTs were analyzed separately from cohort studies. We failed to pool consistent results from RCTs and cohort studies in this study. Although some RCTs examining the effectiveness of ICSI in non-male factor infertility patients, results from currently available cohort studies indicate there is no advantage of ICSI over IVF as an insemination method for non-male factor infertility [25]. A Cochrane review including only one randomized control study even demonstrated IVF had a benefit in fertilization rate than ICSI [50]. Our results indicate that there were no statistically significant differences in fertilization rate (RR = 1.16, 95% CI: 0.83–1.62) between two groups in 4 RCTs; however, a statistically significant higher fertilization rate (RR = 1.16, 95% CI: 1.03–1.31) in the ICSI group was observed in cohort studies. Significant heterogeneities were observed in both subgroups (I² = 97%, P < 0.01, I² = 94%, P < 0.01). The previous prospective clinical trial suggested that the fertilization rate reduced consequently with maternal age increased [51]; the heterogeneity may come from age, racial, and ethnic differences; different protocols of controlled ovarian hyperstimulation (COH); different supplements in culture medium; different duration of studies; and different experimental designs. Moreover, both the results from RCTs and cohort studies were not robust as previously mentioned. It suggests that there are important and potential bias factors related to the effect, and the results must be interpreted cautiously.

Contrary to popular belief, our results indicated that IVF may be beneficial to improve clinical outcomes in non-male factor infertility. Both clinical pregnancy rate (RR = 0.85, 95% CI: 0.77–0.94) and implantation rate (RR = 0.78, 95% CI: 0.65–0.95) were significantly higher in the IVF group in cohort studies. Further subgroup analysis according to the female age demonstrated that ICSI had disadvantages in both CPR and IR among females aged 35 and over (RR = 0.73, 95% CI: 0.59–0.90; RR = 0.52, 95% CI: 0.36–0.74). Nevertheless, the disadvantage was not observed among women aged under 35 (RR = 0.87, 95% CI: 0.71–1.08; RR = 0.93, 95% CI: 0.77–1.12). Moreover, women with age data not available were included in the third subgroup; high heterogeneities were observed in these data sets (I² = 75%, P = 0.02; I² = 82%, P = 0.02). Therefore, we have reason to suspect that age was probably a major reason for existing heterogeneity and may be an important factor that affecting CPR and IR between these two groups. In spite of there was a study favored the use of ICSI in the older IVF population [52], the findings of our study are consistent with the results of previous studies showing that patients with advanced age presented declined pregnancy rate and implantation rate [53–55]. ICSI may yield worse reproductive outcomes in terms of pregnancy rate and implantation rate when the patient’s age is advanced. Sunderam et.al suggested a similar result that advanced age alone may not be a good indication for the use of ICSI, although it was increasingly implemented in older women worldwide [17, 56]. In addition, it suggested that advanced paternal age was associated with a significant increase in the prevalence of both genomic and epigenomic sperm defects [57]. Endometrial receptivity and embryo quality are two critical factors that affect the implantation rate and pregnancy rate. Conventional ICSI is associated with mechanical damage to the oocyte during the conventional ICSI procedure; laser-assisted ICSI is introduced to overcome this problem [58]. A previous study showed that it led to improved embryonic development and clinical outcomes; moreover, the pregnancy rate was increased in older patients especially [59]. In our results, ICSI was associated with lower CPR and IR among female aged 35 and over; these results might be attributed to that oocytes from women with advanced age maybe more venerable to mechanical damage, leading to impaired embryonic development and disadvantageous clinical outcomes.

A clinician may split sibling oocytes into different insemination methods or assign eligible couples in IVF or ICSI treatment cycle when conducted studies on this subject. In the former study design, the embryos are often mixed together to transfer in order to obtain a better clinical effect. It is not possible to extra data like clinical pregnancy rate, implantation rate, and other data that were based on the individual patient. In addition, rescue-ICSI may be applicable for unfertilized oocytes in conventional IVF to prevent total fertilization failure and that may have an effect on clinical outcomes, even though the authors did not mention that in their studies. More importantly, the patient groups are often small; it is difficult to conduct an RCT to assign eligible couples in IVF or ICSI treatment due to ethical considerations. Although more and more scholars pay attention to the effectiveness of ICSI in non-male factor infertility, only 15 studies (2 RCTs and 13 cohort studies) and 9 (1 RCTs and 8 cohort studies) of all included studies provided data of CPR and IR. Therefore, more well-designed RCTs are required for further evaluating the indications of ICSI in non-male factor patients.

Secondary outcomes

Pooled results from cohort studies demonstrated LBR was significantly higher in the IVF group (RR = 0.86, 95% CI: 0.79–0.94). High heterogeneity was noted among these studies (I² = 91%, P < 0.01). Further subgroup analysis according to female age (< 35, ≥ 35) demonstrated that not only was the LBR significantly higher in the IVF group (RR = 0.93, 95% CI: 0.92–0.94; RR = 0.86, 95% CI: 0.78–0.83) but also the heterogeneities were not detected in both groups (I² = 0%, P = 0.54; I² = 0%, P = 0.49). It indicated that age was the main factor of existing heterogeneity. We failed to draw consistent results from the single RCT which suggested there was no difference between two groups regarding live birth rate in the treatment of non-male factor infertility (RR = 1.17, 95% CI: 0.43–3.15). This finding is relatively consistent with the previous meta-analysis based on 4 RCTs [60]. It is undeniable the design and methodological differences existing between RCTs and cohort studies may be a potential cause of the divergence.

Our results revealed MR was higher in the ICSI group than the IVF group (pooled RR = 1.04, 95% CI: 1.01–1.06). Chromosome abnormality, thrombophilia, metabolic disorders, anatomical causes, and immune factor are all important factors that can give rise to miscarriage. To our knowledge, this is the first systematic review and meta-analysis to investigate MR in ICSI treatment of non-male factor infertility. Unfortunately, the result was not robust as previously mentioned that five of the included studies could influence the pooled results. As mentioned before, there were only 7 cohort studies were included in this part; no RCTs reported the miscarriage rate between two group; it is difficult to conduct an RCT to assign eligible couples into two treatment groups due to ethical and clinical effects; further well-designed, multi-center RCTs regarding the effect of ICSI for patients without male factor are warranted.

Limitations

The main limitation of this systematic review and meta-analysis was the different study designs of the included studies. In addition, we failed to draw consistent results between cohort studies and RCTs; the results may be due to systematic differences rather than intervention effects. In addition, there was publication bias in the studies providing FR data; these may due to omitting studies or unpublished literature; besides, studies pooling results positively tended to be published. Therefore, the results should be interpreted with caution. More well-designed RCTs and robust data are warranted to confirm the findings because the primarily included studies are observational studies.

Conclusions

In summary, the results of our meta-analysis did not show ICSI yields a better clinical outcome than IVF in couples with non-male factor. At present, there is little evidence to support the routine use of ICSI as a treatment for non-male factor infertility. Further researches especially well-designed RCTs are needed to investigate whether ICSI can improve outcomes according to variant factors like age and etiology of infertility.

Authors’ contributions

T.G.: study design, data collection and analysis, drafted and revised the manuscript; L.C.: data collection and analysis, drafted and revised the manuscript; C.G.; data collection and analysis, revised the manuscript; Y.Z.; revised the manuscript and final approval of the version to be published.

Data availability

All data included in this study are available upon request by contact with the corresponding author.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Code availability

None.