Benign prostate hyperplasia (BPH), prostate cancer (PCa), and bladder cancer (BC) are common diseases in the elderly male population. The pathological mechanism of these diseases is not yet fully understood. Inflammation of the prostate, which can cause proliferation of epithelium and stroma, is considered to be related to both BPH and PCa1,2. In addition, urinary tract infection (UTI) is significantly associated with genitourinary cancers (GUC), including kidney, prostate, and bladder cancers3. Trichomonas vaginalis infection is one of the most common sexually transmitted infections (STIs), accounting for approximately 276.4 million new cases annually4. Because most male patients are asymptomatic and remain undiagnosed and untreated, persistent infection may cause chronic inflammation, which may increase the risk of GUC. There is a lack of research into the relationship between T. vaginalis infection and BC; however, some studies have mentioned that T. vaginalis infection may induce proliferation of prostatic epithelial cells and stromal cells5,6. Some in vitro studies showed that PCa may be associated with the up-regulation of the expression of genes that can control cell apoptosis or be overexpressed as a proto-oncogene7,8. The study from Vienna General Hospital discovered that 29/86 (33.7%) patients with BPH were positive for T. vaginalis on polymerase chain reaction (PCR) testing9. The Health Professionals Follow-up Study (HPFS) demonstrated that T. vaginalis seropositivity had a positive correlation with PCa risk10. However, conflicting results have also been reported. Miguelle et al. demonstrated that there was no significant association between T. vaginalis infection and PCa in Caucasian or African-American groups11. Another multicenter study in the USA revealed that patients with a history of STIs and positive STI serologies demonstrated no association with BPH12. In addition, there is still a lack of related literature regarding BC and Asian male populations. Thus, this study aimed to examine the association between T. vaginalis infection and BPH, BC, or PCa.

Materials and methods

Data source

We designed a population-based nationwide nested case–control study and obtained inpatient and outpatient files from Taiwan’s National Health Insurance Research Database (NHIRD). The data were collected from the Longitudinal Health Insurance Database 2005 (LHID2005), a part of NHIRD. We randomly selected approximately 2,000,000 people among the total population. All personal information was encrypted by National Health Research Institutes before released.

Ethical approval

Our study was approved by the Institutional Review Board of Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan (TSGHIRB No.: B-109-31). All stages of the study were carried out in accordance with relevant guidelines and regulations. Because the patient identifiers were encrypted before their data were used for research purposes to protect confidentiality, the requirement for written informed consent from patients for data linkage was waived by Institutional Review Board of Tri-Service General Hospital, National Defense Medical Center, Taipei.

Identification of the case and control groups

We selected patients from 2000 to 2015 who had been diagnosed with BPH, PCa, or BC based on the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes as the case group (Table S1). We defined the date of the first disease diagnosis as the index date. We also used ICD-9-CM codes to identify patients with T. vaginalis infection (Table S1). In contrast, the control groups were patients without BPH, PCa, or BC. Among all patients in the case and control groups, we not only selected patients in a 1:3 case:control ratio, matching based on age and index date, but also excluded (1) women and patients of unknown sex, (2) patient’s aged less than 18 years, and (3) those last diagnosed with trichomoniasis within 1 year before the index date (Fig. 1). The matching method was taken propensity score matching, wherein match tolerance was set at 0.15. The propensity score matching was set as using logistic regression in estimation algorithm and nearest neighbor matching in matching algorithm. The options for nearest neighbor were random in matching order, non-replacement, 1 to 3 matching, and no caliper. The comorbidities in our study included hypertension, myocardial infarction, congestive heart failure, cerebral or peripheral vascular disease, dementia, chronic obstructive pulmonary disease (COPD), type 2 diabetes, renal disease, and malignant disease except PCa and BC. We also evaluated depression as one of the comorbidities in our study because it may be associated with some cancers13.

Figure 1
figure 1

The flowchart of the study design (nested case–control study) from National Health Insurance Research Database in Taiwan.

Covariates for analysis

The covariates in our study included age group (18–44, 45–64, ≥ 65 years), four seasons (spring, summer, autumn and winter), with or without diagnosis of depression, geographical area of residence (north, center, south, east and outlying islands of Taiwan), urbanization level of residence (levels 1 to 4), levels of hospitals as medical centers, regional and local hospitals, and monthly income (in New Taiwan Dollars [NT$]; < 18,000, 18,000–34,999, ≥ 35,000). The urbanization level of residence was defined according to the population, along with various indicators of the level of political, economic, cultural, and metropolitan development. Level 1 was defined as a population of > 1,250,000, and a specific designation as political, economic, cultural, and metropolitan development. Level 2 was defined as a population between 500,000 and 1,249,999, and as playing an important role in the political system, economy, and culture. Urbanization levels 3 and 4 were defined as a population between 149,999 and 499,999, and < 149,999, respectively.

Statistical analysis

The statistical analyses were performed using SPSS version 22.0 (IBM Corp, Armonk, NY, USA). A P-value < 0.05 was considered significant. The Chi-squared or Fisher exact test was used to evaluate distributions between the case and control groups. Continuous variables were evaluated using the t test. Unconditional multiple logistic regression analyses were performed to evaluate the risks of BPH, PCa, and BC associated with trichomoniasis after adjusting for age, insurance premium, comorbidities, season, urbanization, and level of care. Adjusted models with significant covariates were constructed using background selection with the likelihood ratio test.


Demographic characteristics of the study population

Table 1 demonstrates the population distribution of different characteristics for 62,544 cases and 187,632 controls from 2000 to 2015. There were no significant differences in age between groups after matching. The proportion with trichomoniasis in the case group was 0.02% (14/62,544), while it was 0.01% (14/187,632) in the control group (P < 0.001).

Table 1 Characteristics of the study group.

Variable evaluation in the multiple logistic regression

We present the results of the multivariable logistic regression analyses in Table 2. Patients with trichomoniasis had a significantly higher risk of BPH, PCa, or BC (adjusted odds ratio [AOR] = 2.999, 95% confidence interval [CI] = 1.426–5.301, P = 0.002). There was also a significantly higher risk for patients with depression (AOR = 3.124, 95% CI = 1.808–4.838, P < 0.001). The opposite result was noted in patients with middle or high insurance premiums (insurance premium NT$18,000–34,999: AOR = 0.745, 95% CI = 0.688–0.799, P < 0.001; insurance premium > NT$35,000: AOR = 0.836, 95% CI = 0.701–0.979, P = 0.019). Patients diagnosed in summer, autumn, or winter also had significantly lower risk than the control group (summer: AOR = 0.938, 95% CI = 0.902–0.953, P < 0.001; autumn: AOR = 0.790, 95% CI = 0.758–0.805, P < 0.001; winter: AOR = 0.862, 95% CI = 0.824–0.878, P < 0.001). Patients who lived in areas with a higher urbanization level had a significantly higher risk of BPH, PCa, or BC (urbanization level 1: AOR = 1.160, 95% CI = 1.124–1.189, P < 0.001; urbanization level 2: AOR = 1.211, 95% CI = 1.179–1.235, P < 0.001) but had significantly lower risk when diagnosed at a higher level of care (hospital center: AOR = 0.819, 95% CI = 0.796–0.902, P < 0.001; regional hospital: AOR = 0.745, 95% CI = 0.724–0.808, P < 0.001) instead.

Table 2 Risk of BPH/prostate cancer and bladder cancer based on stated variables analyzed using multivariable logistic regression.

Risk of BPH/PCa and BC in the trichomoniasis group stratified by covariates

The risk of BPH, PCa, or BC stratified based on variables using multivariable logistic regression is shown in Table 3. Patients with trichomoniasis had a 2.999 times higher risk of BPH, PCa, or BC than the control group (AOR = 2.999, 95% CI = 1.426–5.301). In the case of trichomoniasis, there were significantly higher risks of BPH, PCa, or BC in patients aged > 65 years old, with lower insurance premiums, with/without depression, first diagnosed in winter, urbanization level 2, and first diagnosed in a local hospital (age > 65 years: AOR = 3.685, 95% CI = 1.704–8.015; insurance premium < NT$18,000: AOR = 2.999, 95% CI = 1.326–5.301; with depression: AOR = 3.104, 95% CI = 1.706–5.972; without depression: AOR = 2.545, 95% CI = 1.138–4.289; first diagnosed in winter: AOR = 4.806, 95% CI = 1.104–19.675; urbanization level 2: AOR = 3.284, 95% CI = 1.057–10.978; first diagnosed in local hospital: AOR = 15.121, 95% CI = 1.762–118.976).

Table 3 Risk of BPH/prostate cancer and bladder cancer stratified by variables listed in the table by using multivariable logistic regression.

Risk of BPH/PCa and BC in subgroup with T. vaginalis exposure and the joint effect

Table 4 presents the T. vaginalis exposure ratio in each subgroup of BPH/PCa and BC. T. vaginalis exposure is significantly associated with a higher risk of BPH and PCa (BPH: AOR = 2.685, 95% CI = 1.233–4.286, P = 0.013; PCa: AOR = 5.801, 95% CI = 1.296–26.035, P = 0.016), but has no significant association with BC (AOR = 4.012, 95% CI = 0.524–31.145, P = 0.151). In addition, patients with both depression and T. vaginalis exposure had a significantly higher risk of developing BPH, PCa, or BC in comparison with other groups with only one condition or without them (AOR = 7.682, 95% CI = 5.730–9.451, P < 0.001) (Table 5, Fig. 2).

Table 4 BPH/prostate cancer and bladder cancer subgroups analyzed using multivariable logistic regression.
Table 5 Risk of BPH/prostate cancer or bladder cancer stratified by trichomoniasis and depression status using logistic regression.
Figure 2
figure 2

Risk of BPH/prostate or bladder cancer stratified by trichomoniasis and depression status using logistic regression.


We designed this case–control study based on nationwide data from Taiwan NHIRD. We found that T. vaginalis infection was significantly associated with BPH and PCa in a male population. Therefore, T. vaginalis could be a pathogen that induces BPH and PCa. However, there was no significant association between trichomoniasis and BC. Furthermore, patients with both trichomoniasis and depression had 7.682 times higher risk of developing BPH, PCa, or BC. This result suggests that the joint effect of trichomoniasis and depression could increase the risk of BPH, PCa, or BC.

The mechanism of T. vaginalis inducing BPH and PCa still remains unclear. Several studies have demonstrated different possible mechanisms. In women, T. vaginalis induces pro-inflammatory cytokine production, including interleukin-6 (IL-6), interleukin-8 (IL-8), and chemokine ligand 2 (CCL2), while attaching to vaginal epithelial cells14. A similar inflammatory reaction was also noted in T. vaginalis-infected prostatic epithelial cells in some in vitro studies5,6. Repeated cell damage and repair in chronic inflammation is likely to play an important role in inducing BPH15. Furthermore, the alteration in cytokine expression during chronic inflammation may have effects on cell growth and proliferation of the prostate epithelium and stroma in BPH15. The activated mast cells stimulated by T. vaginalis-infected prostatic epithelial cells can initiate IL-8 and CCL2 expression5. IL-8 could be a predictive marker for BPH16. Some in vitro studies demonstrated that IL-8 can stimulate fibroblast growth factor 2 (FGF-2), which causes the mitosis of prostate stromal cells17. IL-8 could also cause cyclin D1 expression to promote stromal cells proliferation18. In addition, CCL2, secreted by the prostatic stroma fibroblast, could promote both BPH and PCa progression5.

T. vaginalis possibly induces carcinogenesis of the prostate. The infected prostatic epithelial cells produce IL-6 in chronic inflammation19. In early studies, an elevated serum IL-6 level was noted in patients with advanced PCa20. The positive correlation between IL-6 receptor expression and cell proliferation has been reported21. IL-6 also induces epithelial–mesenchymal transition (EMT) in breast cancer growth and metastasis22, and the same reaction may also occur in prostatic epithelial cells23. In addition, more than one study has demonstrated that IL-6 could enhance androgen receptor (AR) activity and AR gene expression24, which is also related to prostate cancer growth. Twu et al. demonstrated that T. vaginalis macrophage migration inhibitory factor (TvMIF) plays an important role in inducing PCa7. There are already studies that have proven that higher human macrophage migration inhibitory factor (HuMIF) levels are present in several cancers, including PCa25. The structure of TvMIF is similar to that of HuMIF, which might explain why TvMIF also has the ability to promote cell proliferation, sustain inflammation, and stimulate the growth of prostate cancer cells7.

In previous studies, T. vaginalis could play an important role as a carcinogen of female cervical cancer26,27. However, there is no consensus regarding the relationship between trichomoniasis and cervical cancer28. Likewise, the role of T. vaginalis in the development of PCa is still controversial. Zhu et al. demonstrated that there was a negative association between PCa and trichomoniasis29. Instead, they discovered culture supernatant of T. vaginalis not only inhibited growth but also induced apoptosis of prostate cancer cell. T. vaginalis could enhance anti-proliferative molecules and decrease the expression of anti-apoptotic molecule29. The T. vaginalis adhesion protein could induce T helper 2 cell cytokines reaction to stimulate the productions of specific antibody30. This enhancement of the immune response might suppress the cancer cell activity31. Moreover, another further study also showed that T. vaginalis seropositivity does not raise mortality risk in men with PCa32. The inflammatory response caused by T. vaginalis might not have influence in the development and progression of PCa32. However, the detail mechanism of immune response between T. vaginalis and prostate epithelial cell still remained unclear, further investigations are necessary.

There were still a lack of studies to prove that trichomoniasis is associated with BC. We still included BC patients in our study because the inflammatory cytokines found in trichomoniasis, including IL-6 and IL-8, are also associated with a higher risk of developing BC33,34 and some parasites, such as Schistosoma haematobium, can induce BC. However, our study shows no significant association between T. vaginalis infection and BC probably because of limited sample.

We added depression as one of the comorbidities in our study due to another previous nationwide population-based cohort study in Taiwan which showed that patients with trichomoniasis had higher risks for developing an individual psychiatric disorder, including depression, anxiety, bipolar disorder, schizophrenia and substance abuse35. Our study results demonstrate that except for depression, no comorbidities had a significant association with BPH, PCa, or BC. The joint effect of trichomoniasis and depression increased the risk by 7.682 times that of the control group. A recent study showed that depression is associated with decreased immunity36. Moreover, depression can also cause cytokine dysregulation and increased serum IL-6 concentration36, which might enhance carcinogenesis after T. vaginalis infection.

Although this study was a large-scale population-based nationwide design with long-term monitoring from 2000 to 2015, there are still several limitations. First, the NHIRD does not contain detailed information regarding the symptom severity of BPH, the histological and TNM classification of PCa and BC, serum sex hormone concentrations, Prostate-Specific Antigen (PSA) levels, T. vaginalis antibody test, family history, or personal history such as sexual exposure, physical activity, alcohol consumption or tobacco smoking. Second, we did not include body mass index (BMI) as one of our variables. Obesity is one of the risk factors for BPH and PCa37, which might affect their association with trichomoniasis. Third, our study might underestimate the exact number of patients with trichomoniasis. Most male patients would not seek treatment due to being asymptomatic, and ineffective screening protocols because of the lack of public health awareness could also lead to possible T. vaginalis infection being neglected38. Another reason that caused underestimation of our case group is that the antibody tests of T. vaginalis were not performed popularly during diagnosis and mostly were female patients39. It is possible that T. vaginalis was substantially undercoded and underrepresented in the study population. Fourth, the number of cases of BC might be too small to be significant and the tracking time might not be sufficient for disease monitoring. Trichomoniasis can be a chronic infection. The outcomes in the study might present later in life, so in some men trichomonas exposure may happen a few years before these outcomes appear or many decades prior to diagnosis. Fifth, the outcome of each case was defined as the first code for BPH, PCa, or BC. This assumes that there is a common pathway between trichomoniasis and these 3 separate diseases. However, this approach method could also ignore these outcomes as comorbidities. For example, patients with PCa or BC could also have BPH or other urinary symptoms. It is possible that many PCa or BC outcomes were ignored if BPH was coded first. This might be another reason that our study samples were underestimated. Sixth, our study was designed as an observational case–control study, so the causation cannot be detected. We hope that in the future more research will support our thesis.


Male patients with T. vaginalis infection have an increased risk of developing BPH and PCa, especially in trichomoniasis patients with comorbid depression. Due to the lack of awareness of this pathogen, clinicians should not only treat patients who are already diagnosed but should also pay more attention to groups with higher trichomoniasis exposure risk.