Is serum PSA a predictor of male hypogonadism? Testing the hypothesis

ABSTRACT Objective: Male hypogonadism (MH) is common among infertile men. Besides testosterone, limited MH biomarkers are available, while researchers have suggested the use of prostate-specific antigen (PSA) to help diagnose MH. Hence, we sought to evaluate the potential use of PSA to predict MH among relatively young men with infertility in Nigeria. Materials and methods: The study included 707 male partners (35–44 years) in infertile couples seeking infertility evaluation at a third-level care center in Nigeria. MH was diagnosed using standard guidelines. Receiver operating characteristic (ROC) and regression analyses explored the potential of serum free PSA (fPSA) and total PSA (tPSA) in predicting MH and MH-related clinical features. Results: In all, 29.7% of the patients had MH (MH+ve). The MH+ve group had lower mean values of fPSA and tPSA than the group without MH (MH-ve). The best fPSA threshold of < 0.25 μg/L compared with the best tPSA threshold of < 0.74 μg/L had higher accuracy (area under the curve [AUC] 0.908 versus 0.866, respectively), sensitivity (87% versus 83%, respectively), and specificity (42% versus 37%, respectively) for MH diagnosis. After adjustment for confounders, fPSA level ≤ 0.25 μg/L was more likely to predict MH-related decreased libido (odds ratio [OR] 2.728, p<0.001) and erectile dysfunction (OR 3.925, p<0.001) compared with tPSA ≤ 0.74 μg/L in the MH+ve group. Conclusion: For MH diagnosis, fPSA and tPSA had good sensitivity but very poor specificity, although fPSA had better potential for MH diagnosis and association with MH-related clinical features than tPSA. Hence, fPSA could complement other biomarkers for MH diagnosis in men 35–44 years, although we recommend further studies to confirm these findings.


INTRODUCTION
M ale hypogonadism (MH) is a spectrum of clinical syndromes associated with diminished testicular function due to testosterone deficiency or impaired action (1,2). Characteristic findings of MH include semen abnormalities, sexual dysfunction, infertility, metabolic changes, and psychosocial abnormalities (2)(3)(4). The etiology of MH may be primary, secondary, congenital, or acquired (1)(2)(3). The occurrence of MH is common among men presenting with infertility (2,5). In northern Nigeria, a 45.9% prevalence of MH has been documented among men seeking infertility evaluation (2,5). Evidence indicates that MH is associated with various other noncommunicable disorders among men (6).
MH is a major endocrine disorder in men, regardless of race, socioeconomic status, or geographical location (4)(5)(6). Men of all ages can be affected, and the clinical presentation depends on the timing of puberty onset (2)(3)(4)(5). Testosterone synthesis tends to decline after the age of 40 years at a yearly rate of 0.4-2% in men (1,2). This decline has been linked to a predominance of MH in older individuals (3)(4)(5)(6). Even though MH is common among elderly individuals, epidemiologic evidence also indicates it as having a high incidence in younger men (5)(6)(7), although the clinical manifestations of MH are less pronounced in these individuals compared with older men (7).
The parameter most widely used for MH diagnosis remains the accurate laboratory determination of decreased serum testosterone levels with concomitant MH clinical features (4). Other biomarkers have been suggested, but they have yet to gain popularity in clinical practice (8). Serum testosterone determination is highly subject to various preanalytical and biological PSA, a predictor of male hypogonadism Arch Endocrinol Metab. 2021;65/2 factors, especially among older individuals. These factors have resulted in a search for other biomarkers to help establish the diagnosis of MH (8,9).
Testosterone exerts multisystemic functions in men (10). The most pronounced effect is on the prostate, where testosterone influences the expression of the prostate-specific antigen (PSA) gene initiating PSA synthesis (11). The influence of testosterone on PSA synthesis has prompted the interest of some investigators suggesting that PSA is a biomarker of MH (12). However, most of these investigators had evaluated PSA levels mainly among middle-aged and older men, who are likely to have prostate disorders and other comorbid conditions that could affect the association between MH and PSA (12)(13)(14).
The association between PSA and testosterone is less variable in younger men, and a positive correlation between PSA and testosterone levels has been consistently established in younger compared with older men (13). Additionally, the "saturation" theory suggests that the relationship is mainly evident at low testosterone concentration, which may favor MH diagnosis (15).
Hence, the present study aimed to evaluate the predictive potentials of serum total PSA (tPSA) and free PSA (fPSA) on MH and MH-related features among young men of Nigerian origin presenting for infertility evaluation.

Study location and design
This prospective, descriptive, and cross-sectional survey was conducted at the University of Port Harcourt Teaching Hospital (UPTH) in Nigeria. UPTH is a third-level health care facility located along the famous East-West Road in Nigeria. The hospital is a major referral center in the southern part of the country and has various specialized departments.

Ethical considerations
The study protocol was approved by the Institutional Research Ethics Committee (approval reference UPTH/ADM/90/S.11/VOL.XI/851). Informed consent was submitted by all subjects upon enrollment. The study was conducted according to the UPTH research ethics standards and the Declaration of Helsinki.

Sample size determination and sampling protocol
The minimum sample size required for the study was 377, which was computed using the formula n = Z 2 pq/d 2 (n = sample size, Z = standard normal deviation -set at 1.96, p = proportion in the target population estimated to have the particular characteristics of interest [MH]), using 45.9% as reported within the study region, q = 1 -p (proportion in the target population not having the characteristics of interest, d = degree of accuracy required, set at 0.05) with a 95% confidence interval (CI) and a power of 80% (5,16). To enhance the study power, we recruited 707 individuals.
Convenience sampling was applied to recruit the participants from March 2018 to January 2020.

Study population
The study population comprised male partners in infertile couples recruited from the Departments of Obstetrics-Gynecology and Chemical Pathology outpatient units. In these units, male partners in infertile couples usually undergo endocrine evaluation following abnormal seminogram findings.
Other criteria for exclusion were lack of consent, presence of any chronic debilitating illness, prior prostatectomy and orchiectomy, and use of medications PSA, a predictor of male hypogonadism Arch Endocrinol Metab. 2021;65/2 known to influence serum PSA levels such as statins, nonsteroidal antiinflammatory agents, thiazide, calcium supplement, aspirin, 5-alpha-reductase inhibitors, alpha-blockers, other anti-androgens, and exogenous testosterone.

Acquisition of non-laboratory data
Data on clinical and anthropometric variables obtained included age, systolic blood pressure (SBP, mmHg), diastolic blood pressures (DBP, mmHg), weight (kg), height (m), body mass index (BMI) calculated with the formula weight (kg)/height (m 2 ), and waist circumference (cm). Prostate volume (cm 3 ) was calculated using the prolate ellipsoid formula with dimensions obtained from TRUS of the prostate (18). The IPPIS and Androgen Deficiency in the Aging Male (ADAM) questionnaires were also administered (17,19).
While the IPPIS questionnaire evaluated urinary symptoms over the preceding 4 weeks due to prostate enlargement (with 0-7 denoting no or mild symptoms), the ADAM questionnaire was a validated screening tool for identifying men at risk of MH (17,19).
The ADAM questionnaire consisted of 10 questions with "yes" or "no" answers concerning MH symptoms. A positive ADAM questionnaire response is a "yes" answer to either question 1 (reduced libido) or 7 (erectile dysfunction) or any three other questions (19).

Acquisition of laboratory data
Seminogram was carried out in the center using standard protocols (20). The parameters determined were sperm count (normal >15 x 10 9 /mL), sperm morphology (normal > 4%), sperm total motility (normal > 40%), and semen volume (normal 1.5 mL), as defined by the World Health Organization (WHO) (20). Identified cases with abnormal seminogram results were usually confirmed on repeat analysis within 2-4 weeks.
A total of 8 mL of 10-12-hour fasting venous blood was drawn between 8-10 am from each participant's antecubital vein. Of that, 5 mL were added to tubes without anticoagulant and allowed to clot undisturbed for 1 hour for full retraction, then centrifuged at 2500 revolutions per minute. The serum supernatant was separated with a Pasteur's pipette into tubes without anticoagulant and frozen at -80 o C until analysis of tPSA, fPSA, serum follicle-stimulating hormone (sFSH), serum luteinizing hormone (sLH), serum prolactin (sPRL), and serum total testosterone (sTT).
The remaining 3 mL of blood drawn were added to an ethylene diamine tetra-acetic acid tube and centrifuged immediately. The supernatant plasma was separated and stored frozen until analysis of plasma sex-hormonebinding globulin (SHBG) and albumin.
Following analysis, samples with sTT in the hypogonadal range were identified, and the patient was contacted for repeat sampling and laboratory analysis within 2-4 weeks for confirmation, based on Endocrine Society guidelines (4).
Quality control specimens (Monobind Incorporated) with known hormone concentrations spanning higher and lower concentrations were used to determine analytical precision.

Data management and statistical analysis
The acquired data was entered into the software SPSS (version 21.0; SPSS Inc. Chicago, IL, USA) and subsequently analyzed. Continuous data were expressed as mean ± standard deviations and compared with independent Student's t test. Categorical data were expressed using numbers and percentages and compared with the chi-square test. The diagnostic potentials and the optimal cutoff values of tPSA and fPSA to predict MH were determined using receiver operating characteristics (ROC) curve analysis. Binary logistics regression analysis was used to evaluate relationships between variables. Statistical differences were considered significant at p < 0.05.

RESULTS
During the study period, 2,820 infertile couples presented for infertility evaluation in the study center. Applying the eligibility criteria, a total of 768 male partners in infertile couples met the eligibility criteria and were enrolled. However, 61 were further excluded due to failure to present for blood or semen sampling (n = 21), inability to present for repeat blood sampling (n = 15), inability to present for repeat seminogram (n = 15), and abnormal DRE and TRUS finding (n = 10), yielding a final population sample of 707 patients. Among these, MH+ve was documented in 210 (29.7%) patients, while 497 (70.3%) were categorized as MH-ve.

DISCUSSION
The occurrence of MH is common among infertile patients (5,24). Geidem and cols. evaluated the hormonal profile of men investigated for infertility in northern Nigeria and reported a 45.9% prevalence of MH in their study cohort (5). Although this prevalence rate is higher than the one documented in the present study (29.7%), this could be due to differences in study designs. We studied men aged 35-44 years and used a set of criteria for MH diagnosis, while Geidem and cols. evaluated men aged 22-52 years using only a single criterion for MH diagnosis. Other higher and lower variations in MH frequency have been documented in the literature (24). Apart from sTT, there are limited numbers of biomarkers described in the literature to predict MH accurately (8). Some other biomarkers, such as insulinlike factor 3, have been suggested and evaluated but have yet to gain popularity in clinical practice (25). The PSA gene is among the genes positively influenced by sTT (11). Among men with suboptimal sTT status, a positive correlation between PSA and sTT has been documented (13,15). At optimal sTT status, the androgen receptor is saturated, and further rises in sPSA in normal sTT status are independent of the sTT influence (15). This relationship has prompted few studies geared towards the evaluation of sPSA properties over MH diagnosis (12,13).
We evaluated the diagnostic properties of tPSA and fPSA over MH among relatively young men and demonstrated that both molecular PSA forms predict MH+ve significantly with good sensitivity but very poor specificity, thereby making both molecules poor diagnostic parameters for MH+ve diagnosis among young men. Compared with the strict diagnostic criteria for MH diagnosis applied in the present study, tPSA showed good diagnostic properties in predicting MH+ve, although better properties were established by fPSA.
At the best sensitivity and specificity to perfectly predict MH by fPSA, obtained at < 0.25 µg/L, this marker demonstrated better sensitivity, specificity, PPV, and NPP than that rendered by the best tPSA cutoff threshold of < 0.74 µg/L. Our findings are in line with a closely similar diagnostic study by Rastrelli and cols., who evaluated the diagnostic properties of tPSA (but not of fPSA) on MH diagnosis and documented observations, although with slightly different diagnostic properties than ours (12).
Compared with the diagnostic properties of tPSA demonstrated in the present study, Rastrelli and cols. reported lower accuracy (AUC 0.612 ± 0.001) and sensitivity (65.2%) but higher specificity (55.5%) over MH diagnosis at best tPSA threshold of < 0.65 µg/L (12). However, in contrast to our study design, Rastrelli and cols. evaluated predominantly middle-aged and older individuals, which may have contributed to the variations in diagnostic properties obtained in their study and ours.
When the MH-related sexual domains (erectile dysfunction and decreased libido) of the ADAM questionnaire were evaluated in the present study, a fPSA threshold of ≤ 0.25 µg/L also exhibited a greater likelihood of predicting erectile dysfunction and decreased libido in the MH+ve group compared with a tPSA threshold of ≤ 0.74 µg/L. Our findings compare with those of other studies in the literature reporting low tPSA association with MH-related sexual dysfunction (SD) (12,13).
In a retrospective cross-sectional study of patients presenting with SD, Coronal and cols. observed that subjects with higher tPSA levels experienced lower frequencies of SD (including erectile dysfunction and loss of libido) (13). Rastrelli and cols. had also concluded in their study that low tPSA was associated with impaired sex-and sleep-related erections (12).
The relatively better diagnostic properties rendered by the fPSA over MH diagnosis could be attributed to some factors (26,27). According to some data, fPSA -and not tPSA or age -has a better predictive value of prostate volume, which ultimately determines PSA secretion and serum concentration (24). Levels of fPSA have been documented to be highly sensitive to slight changes in sTT compared with those of tPSA (15,26,27).
The "saturation" hypothesis proposes that PSA is sensitive to TT variations at or below the near-severe hypogonadism level (8 nmol/L, the level applied in the present study) but is insensitive to testosterone variations above this concentration (15). Hence, we hypothesize that the changes inherent in PSA, as proposed by the "saturation" hypothesis, may be more pronounced with fPSA since this is the physiologically active molecular fraction of PSA instead of tPSA.
The study has some limitations worthy of note. First, it was a single-hospital-based survey, and the findings may not represent the entire population within the study region. Second, the study's findings were PSA, a predictor of male hypogonadism Arch Endocrinol Metab. 2021;65/2 obtained from men of black race, which may also not reflect those of Caucasian populations. Third, we relied on calculated serum testosterone parameters, due to limited resources, instead of directly measuring these parameters, although the correlation between calculated and directly measured testosterone parameters has already been established (21). Fourth, being a crosssectional study, the findings do not allow deduction of a cause-and-effect relationship, and the duration of the study does not guarantee stability to the associations observed between variables over different time periods.
In conclusion, MH is common among infertile couples and has limited biomarkers dedicated to its diagnosis. Investigators have previously suggested the usefulness of tPSA in identifying MH, which prompted this study. Although both fPSA and tPSA showed good sensitivity for MH diagnosis, they also had very poor specificity, making both molecules poor diagnostic parameters for MH evaluation in young men. Still, fPSA had better MH diagnostic potential and association with MH-related clinical features than tPSA. Hence, fPSA could complement other biomarkers for MH diagnosis among relatively young men. Further studies are suggested to confirm the findings of the present study.