Introduction

Breast cancer (BC) is the most common form of cancer globally and the most commonly reported cause of cancer-related deaths among women [1]. Advances in the diagnosis and treatment of BC have improved long-term survival of due to which an increasing number of women are now surviving from BC. The 5-year survival rate for BC patients is approximately 76–92% globally [2], and approximately 80% in China [3] after systematic treatment. Female sexual dysfunction (FSD), which includes abnormalities in sexual desire, arousal, lubrication, satisfaction, orgasm, and dyspareunia, is one of the most common complications in women with BC [4].

In 2000, Rosen et al. [4] published the female sexual function index (FSFI), which assesses the abovementioned six sexual function-related domains. FSFI is currently one of the most widely used tools to assess female sexual function. It can be employed to assess the sexual feelings and reactions of subjects during the previous 4 weeks prior to assessment. FSD is usually indicated by an FSFI score < 26.5. The FSFI can be used for self-assessment, and the survey has been used successfully in patients with malignant tumors. The prevalence of FSD [5,6,7,8,9] and the range of sexual function scores in women with BC differ greatly among published studies [5, 7, 10,11,12,13]. To our knowledge, there is no meta-analysis of the studies which estimated the prevalence of FSD and the degree of sexual function in women with BC using the FSFI in literature. This study was therefore designed to estimate FSD prevalence and assess sexual function in women with BC based on the FSFI scores. These results can be used to provide guidance for the clinical care of women with BC, and will offer an objective basis for health decision-making in this growing population.

Materials and methods

This meta-analysis was performed by following the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines.

Literature database search strategy

We searched PubMed, Embase, the Cochrane Library, the Chinese National Knowledge Infrastructure Database (CNKI), the Chinese Wan Fang Database (WanFang Data), and the Chongqing VIP database (VIP), and retrieved Chinese and English FSFI studies examining FSD prevalence and sexual function in women with BC published between April 2000 and January 2017. We also manually searched the references sections of the selected papers to explore additional relevant literature. These searches were performed using a combination of subject words and keywords, which included the phrases “breast neoplasm” and “female sexual function”. The following search string was used in PubMed: (((((((((((((((((((((((((((breast neoplasm) OR neoplasm, breast) OR neoplasms, breast) OR tumors, breast) OR breast tumors) OR breast tumor) OR tumor, breast) OR mammary neoplasms, human) OR human mammary neoplasms) OR neoplasm, human mammary) OR neoplasms, human mammary) OR mammary neoplasm, human) OR mammary carcinoma, human) OR carcinoma, human mammary) OR carcinomas, human mammary) OR human mammary carcinomas) OR mammary carcinomas, human) OR human mammary carcinoma) OR breast cancer) OR cancer, breast) OR cancer of breast) OR mammary cancer) OR malignant neoplasm of breast) OR malignant tumor of breast) OR breast carcinoma) OR cancer of the breast) AND (female sexual function).

Study selection and data extraction

Literature screening, data extraction, and cross-checking were carried out independently by two authors with the assistance of an expert in epidemiology and statistics. Disagreements regarding study selection were resolved in consultation with the fourth author. Relevant variables including name of the first author, year of publication, age, country and continent where the study was performed, type of study design, sample size, FSD prevalence, total FSFI score, intervention/exposure factors, and follow-up time were extracted according to a preformatted data extraction table. Missing data were obtained by contacting the authors of the respective studies. These variables were entered into Excel to create a database. The prevalence of FSD in women with BC was the primary endpoint of the present study.

The inclusion criteria were as follows: study (1) published in English or Chinese language; (2) published between April 2000 and January 2017 as the FSFI was first proposed in April 2000; (3) design was either observational (including cohort studies and cross-sectional studies) or interventional (including randomized controlled trial, and quasi-experimental studies); (4) participants were the female patients diagnosed with BC, regardless of race or disease course; (5) used FSFI to examine sexual function of women with BC during or after BC treatment (surgery, chemotherapy, radiotherapy, or endocrine therapy); and (6) recruited > 30 participants.

The exclusion criteria were as follows: (1) repeat studies with the same sample populations; (2) reviews, case reports, conference abstracts, letters, or comments; (3) the full text was unavailable; (4) the study subjects included non-BC, male, female homosexual, female bisexual, or transgender patients; and (5) the FSFI assessment was applied inappropriately in the study.

Quality assessment

As described by the assessment principles in JBI-MAStARI [14], quality assessment of the included studies was performed independently by the first and second authors, and disagreements were resolved by consensus or with the involvement of the fourth author, if necessary. There were 10 assessment items, with a total score of 0–10 (1 point per item) for quasi-experimental studies, and 9 assessment items with a total score of 0–9 (1 point per item) for cohort studies and cross-sectional studies; 1 was for “Yes” and 0 was for “No” or “Unknown”. Studies with a quality score ≤ 4 were deemed to be of low quality, those with a quality score of 5–6 were deemed to be of medium quality, and those with a quality score ≥ 7 were deemed to be of high quality. All studies with a quality score ≥ 5 were included in this meta-analysis.

Statistical analysis

Meta-analysis of FSD prevalence rate or total FSFI scores was conducted using STATA software version 14.0 by pooling these indices and their 95% confidence intervals (CI) reported by the individual studies under random effects model. The statistical heterogeneity was assessed with I2 and the Q test. I2 < 50% and P > 0.05 (Q test) indicate low/moderate heterogeneity among included studies. Sensitivity analyses were performed by sequentially excluding studies and then observing the robustness of any changes in results. The publication bias of included studies was analyzed with Begg’s rank test and Egger’s regression analysis; P > 0.05 (for both Begg’s rank test and Egger’s regression analysis) indicates the absence of publication bias. Subgroup analyses were performed according to the age, publication year, area (country and continent) where the study was performed, sample size, type of study design, and study quality. Meta-regression analyses were performed using the CMA software version 2.0 to further explore the sources of heterogeneity.

Results

Literature search results

A total of 224 studies were identified, and 19 studies involving 2684 women with BC were ultimately included in this meta-analysis [5,6,7,8,9,10,11,12,13, 15,16,17,18,19,20,21,22,23,24] (Fig. 1). Fifteen of these studies analyzed FSD prevalence in women with BC (n = 2137) and 15 studies analyzed the total average FSFI scores in women with BC (n = 2062). Eleven of the 19 studies analyzed both FSD prevalence and female sexual function measured with FSFI.

Fig. 1
figure 1

Process of study screening and selection

Full size image

Basic characteristics and quality of included studies

Important characteristics of the included studies are presented in Table 1. Four studies were published in Chinese, and 15 studies in English language. The studies were conducted in the following areas: East Asia, 6 studies [5, 6, 10, 15,16,17]; West Asia, 3 studies [8, 9, 12]; North America, 4 studies [11, 18, 19]; South America, 4 studies [11, 18, 19, 22]; Europe, 1 study [22]; Africa, 1 study [23]. Four studies were interventional [7, 9, 18, 19], all with quasi-experimental designs, and the remaining 15 studies were observational, including 3 cohort [6, 8, 15], and 12 cross-sectional studies [5, 10,11,12,13, 16, 17, 20,21,22,23,24].

Table 1 Characteristics and quality of the included studies
Full size table

Fifteen studies reported sufficient data to permit for a meta-analysis of the prevalence of FSD, including 3 quasi-experimental studies [7, 9, 19], 3 cohort studies [6, 8, 15], and 9 cross-sectional studies [5, 10,11,12, 16, 21,22,23,24]. These 15 studies included 2137 women with BC, of whom 1488 experienced FSD. Fifteen studies provided sufficient data to permit for a meta-analysis of total FSFI scores, including 3 quasi-experimental studies [9, 18, 19], 1 cohort study [8], and 11 cross-sectional studies [5, 10,11,12,13, 16, 17, 20, 22,23,24]. These 15 studies included a total of 2062 women with BC. The follow-up time in the quasi-experimental or cohort studies ranged from 30 days to 12 months. Most of the included studies described at least two or more clinical and demographic variables including age, sexual partners, and marital status. The clinical variables included the clinical stage, therapy, and surgery for BC. The quality of all included studies was medium or high.

Meta-analysis results

The heterogeneity test results of the 15 studies analyzing the FSD prevalence in women with BC indicated that there was significant heterogeneity among these studies (I2 = 98.8%, P < 0.01). Therefore, pooling was conducted under random effect model. The FSD prevalence in women with BC was thus found to be 73.4% (95% CI 64.0%, 82.8%) (Fig. 2).

Fig. 2
figure 2

Meta-analysis for the FSD prevalence in women with BC

Full size image

The heterogeneity test results for the 15 studies evaluating the total FSFI scores of sexual function in women with BC indicated that there was significant heterogeneity among these studies (I2 = 97.60%, P < 0.01), so pooling was performed under random effect model. The average FSFI sexual function score in women with BC was 19.28 (95% CI 17.39, 21.16) (Fig. 3).

Fig. 3
figure 3

Meta-analysis for the total FSFI score of sexual function in women with BC

Full size image

Subgroup analysis

To explore the sources of heterogeneity in this meta-analysis, subgroup analyses were performed with regard to the mean age, year of publication, country (mainland China and other countries), continent, sample size, type of study design, and literature quality for each study. The results showed that there were statistically significant regional differences between Asia, America, and Europe in FSD prevalence in women with BC (P < 0.001), and a marginally significant difference between mainland China and other countries (P = 0.06). However, there were no statistically significant differences among different ages, years of publication, sample sizes, types of study design, or literature quality (P > 0.05) (Table 2). There were marginally significant regional differences between Asia, America, and Africa (P = 0.07) and type of study design (P < 0.05) in total FSFI scores. There were no statistically significant differences among different ages, years of publication, countries (mainland China vs other countries), sample sizes, and literature quality (P > 0.05) in total FSFI scores (Table 3).

Table 2 Subgroup analysis for the prevalence of FSD in women with BC
Full size table
Table 3 Subgroup analysis for the total FSFI of sexual function in women with BC
Full size table

Meta-regression analysis

To determine the sources of heterogeneity among the included studies, univariate meta-regression analyses were performed according to the mean age, year of publication, country, continent, sample size, type of study design, and quality of literature. The results revealed that different continents (America vs. Asia) were the main sources of heterogeneity in the meta-analysis of total FSFI scores in women with BC (b = −4.05, 95% CI 7.90, −0.20, P = 0.04). No other factors assessed in this meta-regression analysis were statistically significant.

Sensitivity analysis

For the sake of sensitivity analysis of the 15 studies included in the meta-analysis of the FSD prevalence among women with BC, we successively excluded either 4 studies from mainland China, one study from Europe, 5 studies with sample sizes < 100, and one study with the year of publication < 2012. The results revealed that the FSD prevalence in women with BC was 69.7% (95% CI 56.9%, 82.5%), 74.9% (95% CI 65.3%, 84.5%), 71.4% (95% CI 59.1%, 83.6%), and 73.9% (95% CI 64.1%, 83.7%), respectively, demonstrating no significant differences when compared to the results of overall meta-analysis (73.4%). This indicated that the study results were stable.

For sensitivity analyses of the 15 studies of the total FSFI scores in women with BC, we successively rejected either 3 studies from mainland China, one from Africa, 5 with a sample size < 100, and 3 with a publication year < 2012, and then sensitivity analyses were performed. The results showed that the total FSFI scores in women with BC were 19.53 (95% CI 17.00, 22.07), 19.38 (95% CI 17.42, 21.35), 19.88 (95% CI 17.30, 22.47), and 19.60 (95% CI 17.41, 21.79), respectively, demonstrating no significant differences when compared to the results of overall meta-analysis (19.28). This indicated that the study results were stable.

Assessment of publication bias

In the present study, funnel plots showed no significant asymmetry. Additionally, for both the meta-analysis of FSD prevalence and of the total FSFI scores in women with BC, no publication bias was detected based on Begg’s rank test (Z = 1.09, P = 0.276; Z = 0.20, P = 0.84) or Egger’s regression analysis (t = −0.49, P = 0.629; t = 0.25, P = 0.807).

Discussion

This study is the first FSFI-based meta-analysis of FSD prevalence and the sexual function scores focusing specifically on women with BC. The sample sizes of all included studies were ≥ 30. Sensitivity analyses indicated that the results were robust, and that there was no significant publication bias. The subgroup analyses indicated that neither FSD prevalence nor sexual function scores in women with BC were affected by age, year of publication, sample size, or quality of literature.

In this study, the pooled FSD prevalence in women with BC was 73.4% (95% CI 64.0%, 82.8%), which was greater than the 65.54% (95% CI 46.99%, 84.09%) previously reported by Maiorino et al. [25] This suggests that the FSD prevalence in women with BC is high, with approximately 75% of women with BC developing some degree of FSD. The above difference in FSD prevalence among women with BC may be because Maiorino et al. reviewed literature published through December 31, 2014, did not include Chinese literature databases in literature search, and therefore included 9 studies. In our meta-analysis, we reviewed literature by January 2017 and included 9 studies [6, 7, 9, 10, 15,16,17, 23, 24] published after 2015 of which four [10, 15,16,17] were published in mainland China. Laumann et al. [26] reported that 43% of American female patients developed FSD, whereas approximately 30–63% developed FSD in mainland China [27]. Our meta-analysis results indicate that the prevalence of FSD was higher in women with BC than in non-BC women, suggesting that women with BC constitute a high-risk group. This may be due to their BC-specific treatment experiences, such as body image changes after breast surgery, hormone treatments, changes in hormone levels after ovariectomy, and the physiological and psychological effects of chemoradiotherapy. Given these findings, it is important that greater attention be paid to the sexual function changes affecting women with BC. These results may thereby help to increase the availability of assistance for women with BC who are facing sexual health problems.

Subgroup analysis showed that regarding FSD prevalence in women with BC, there were statistically significant regional differences between Asia, America, and Europe (P < 0.001), and a marginally significant difference between mainland China and other countries (P = 0.06). FSD prevalence in Asian, American, and European women with BC was 74.1% (95% CI 58.3%, 85.5%), 80.4% (95% CI 68.7%, 88.5%), and 52.7% (95% CI 45.5%, 59.8%), respectively. This indicates that the FSD is more prevalent in American women with BC as compared with those in Asia and Europe. These differences may be attributable to differences in ethnic, social, or cultural factors. In our meta-analysis, only one study was included in the European subgroup, therefore additional studies should be conducted to confirm the reliability of these regional differences. FSD prevalence in women with BC was 82.8% (95% CI 77.3%, 87.3%) in mainland China, which was higher than the 72.1% (95% CI 59.7%, 81.8%) observed in other countries, suggesting that Chinese women with BC are at a higher risk of developing FSD. In a survey of 609 non-BC women ≥ 20 years old who underwent a health checkup, Zhang et al. [28] found that FSD prevalence was 56.8% and that it increased with age, which was greater than 43% observed in American women [26]. This suggests that even among non-BC women, FSD prevalence is higher in mainland China than in America. Overall, there are no significant differences in BC diagnosis or treatment between America and China, so the differences in FSD prevalence in women with BC between these regions may instead be related to specific genetic or cultural factors. To date, there has been no epidemiological study comparing the risk factors for FSD in women with BC among different races or individuals from different continents.

Our meta-analysis revealed that total average FSFI score of women with BC was 19.28 (95% CI 17.39, 21.16), which was markedly lower than that in non-BC women (FSFI > 26.5 indicates normal sexual function), [4, 26] but consistent with the 19.58 (95% CI 17.64, 21.53) observed in the previous meta-analysis of Maiorino [25]. This indicates that the overall sexual function in women with BC is significantly worse than in non-BC women. The FSFI is an indicator that is based on sexual function and is affected by several social and psychological factors, with estrogen levels being the most important physiological factor influencing this sexual functionality.

Generally, natural menopause, drugs, or surgical castration can all contribute to reduced estrogen levels in women with BC, and these reduced levels can induce or aggravate FSD, ultimately leading to manifestations of sexual dysfunction such as the absence of sexual desire, reduced sexual arousal, lack of vaginal lubrication, pain during intercourse, difficulty achieving orgasm, and genital hypoesthesia. The age, education, income, surgical procedures, chemoradiotherapy, and hormone treatments undergone by women with BC may all directly or indirectly affect one or more of these symptoms [29, 30] thus reducing their FSFI scores. The quality of one’s sex life is an important dimension of overall quality of life, which is reduced by sexual hypofunction. Therefore, changes in the sexual function in women with BC should be evaluated both at baseline prior to surgery and at different phases during treatment in order to facilitate the early detection and diagnosis of FSD. We also suggest that sexual function-related care should be incorporated into routine care assessments, follow-ups, and interventions among women with BC to improve sexual function and quality of life.

The subgroup analysis showed that there were marginally significant regional differences between Asia, America, and Africa with regard to the sexual function scores of women with BC (P = 0.07), and the pooled total FSFI scores for these regions were 21.40 (95% CI 18.48, 24.32), 17.32 (95% CI 15.15, 19.48), and 17.70 (95% CI 15.55, 19.85), respectively. This suggests that the average sexual function of American women with BC is worse than that of Asian and African women with BC, which may be attributed to a clustering phenomenon. This clustering phenomenon may itself be associated with regional differences in culture, sex education, race, and hospital treatment methods [31,32,33,34]. There were also significant differences among different types of study designs (P < 0.05), which may be related to the literature quality associated with these different types of studies. Cohort studies and quasi-experimental studies are generally of good quality because they comply with the causal sequence and can effectively avoid recall bias. Cross-sectional studies are easily influenced by recall bias and reporting biases and can therefore be of poor quality.

In our meta-analysis, the results showed a high degree of heterogeneity. The subgroup analyses suggest that the sources of heterogeneity include different countries and continents where the studies were performed, as well as the different types of study designs. The meta-regression analysis further indicated that different areas where the studies were performed (America vs. Asia) were the primary sources of heterogeneity. Heterogeneity can also be associated with the following factors: (1) there are no objective standards for the diagnosis and assessment of FSD, and the FSFI is a self-report-based scale with subjective assessment results. (2) Different language translations of the FSFI were used across studies, and investigators interpreted these results privately with unknown scoring methods, potentially resulting in differences in patient comprehension and causing inconsistencies in the reliability and comparability of these results. (3) Some studies failed to completely report on or lacked any report of menopause, sexual partners, marital status, clinical stage, of therapeutic regimen. Therefore, we were unable to perform subgroup analyses based on these clinical and demographic factors to identify other potential sources of heterogeneity. Our subgroup analyses of the ages of subjects, the years of publication, sample sizes, and the quality of literature also failed to reduce this heterogeneity. The heterogeneity in the single-arm meta-analysis was primarily affected by differences in the sample size among the included studies, which is difficult to entirely control for, but it did not have a decisive impact on the accuracy of our study results [35]. Our meta-analysis results are more objective than the results of any one individual study.

This study has the following limitations. (1) Only studies in which the sexual function of women with BC was assessed using FSFI were included. (2) Single-arm observational studies were the dominant type of study design for the included studies. (3) The literature quality of included studies was mediocre. (4) The available studies all had a small sample size. (5) The available literature failed to investigate the severity of FSD more in depth via patient stratification. (6) We also could not conduct subgroup analysis for clinical stage, treatment regimen, and other possible factors that could have affected the outcomes achieved herein.

Conclusion

The regional differences of studies included in our meta-analysis introduced significant heterogeneity. Taken together, our results confirm that women with BC have a high prevalence of FSD and lower average sexual function. Relative to those in Asia, women with BC in America have a higher prevalence of FSD and lower average sexual function. FSD prevalence among women with BC is greater in mainland China than in other countries or areas. To improve the sexual health and quality of life for women with BC, active interventions aimed at treating or preventing FSD should be provided to patients during the diagnosis, treatment, and follow-up of BC.