Haemospermia in the Real- Life Setting: a New High-Risk Stratication

We aimed to validate the EAU guidelines in a homogeneous cohort of men with haemospermia, and to identify a novel and better performing risk stratication compared to EAU guidelines. Data from 283 consecutive patients complaining of a single episode/recurrent haemospermia were retrospectively analysed. Patients were stratied into low vs. high-risk according to EAU guidelines, whose diagnostic performance was then validated. We identied a new risk stratication model based on clinical factors associated with i) positive semen culture and ii) prostate cancer (PCa) and bladder cancer (BC). Diagnostic accuracy of the two predictive models (EAU vs. New) was assessed and decision curve analyses (DCA) tested their clinical benet. Overall, 259 (91.5%) were high-risk and 24 (8.5%) low risk according to the EAU guidelines. Recurrent haemospermia was reported by 134 (47.4%) patients. 126 (44.5%) had baseline CCI score ≥ 1. At MVA logistic regression analysis, history of recurrent genito – urinary tract infections was identied as a predictor for positive semen culture (OR: 3.39, 95% CI: 1.77 – 6.57, p=0.002). Likewise, baseline CCI ≥ 1 was identied as a predictor for PCa and BC (OR: 1.55, 95% CI: 1.17 – 2.04, p=0.009). Sensitivity, specicity, and AUC of the EAU guidelines were 13.3%, 89.2% and 51% respectively, whereas the new model performed substantially better: 98.9%, 58% and 78% respectively. The application of the EAU guidelines risk stratication does not ensure proper identication of high-risk patients complaining of haemospermia. We propose a novel, better performing and easily implementable risk stratication tool.


Introduction
The cause of haemospermia has mostly been ascribed to benign conditions [1][2][3]. Although this holds true, a non-negligible proportion of patients harbour more serious and harmful conditions, which surely deserve further diagnostic tests and investigations [4]. As such, haemospermia may also be regarded as an alarm bell of something that deserves a correct diagnostic interpretation and eventually appropriate treatment [4 -6]. In this context, the identi cation of a proper high-risk category among these patients remains a matter of debate. According to the recently published Sexual and Reproductive Health guidelines of the European Association of Urology (EAU), patients over 40 years of age and patients complaining of recurrent episodes of haemospermia (regardless of their age) are considered "high-risk", thus having a higher likelihood to harbour medical conditions that need to be identi ed and treated (e.g., ongoing semen infection, prostate cancer (PCa), bladder cancer (BC) and testicular cancer (TC)) [1].
These men should undergo a series of diagnostic tests and/or empirical treatments including antibiotics, anti-in ammatory drugs, trans-rectal ultrasound (TRUS), pelvic magnetic resonance (MRI), PCa screening with prostate-speci c antigen (PSA) dosage and digital rectal examination (DRE) (in men ≥ 40 years old), TC screening with testicular ultrasounds (in men <40 years old) and BC screening with cystourethroscopy and eventually biopsy [1,7]. Possibly, a high proportion of patients presenting with haemospermia would fall into the EAU high-risk category undergoing many unnecessary diagnostic and empirical treatments; this, not only translates into higher costs but also provides unneeded information over the patients' diagnostic work-up of these patients [6,[8][9][10]. A proper identi cation of those who would truly bene t from second-level diagnostic investigations is still needed. We aimed to retrospectively validate the EAU guidelines classi cation and to identify a novel risk strati cation to properly identify a new high-risk category of men presenting with haemospermia.

Material And Methods
After institutional review board approval, we retrospectively analysed data from 283 consecutive patients seeking rst medical help for single/recurrent episode of haemospermia at a single outpatient clinic between January 2006 and September 2020. Sociodemographic characteristics, including age and relationship status, were collected for every patient. All subjects were assessed via a detailed medical and drug history. Health signi cant comorbidities were scored using the Charlson comorbidity index (CCI) [11]. The CCI was categorized as 0 or ≥1. Arterial hypertension (HTN) was de ned as o ce systolic blood pressure values ≥ 130 mmHg and/or diastolic blood pressure values ≥ 90 mmHg. Data on recreational habits including physical activity (de ned as at least 2 self-reported physical exercise sessions per week per individual), cigarette smoking history, and alcohol use were also collected. The entire cohort of patients has been longitudinally followed up with outpatient clinical assessments in order to detect the aetiology of haemospermia. Patients lost to rst follow -up evaluation were eventually excluded from the analysis (n=32). Venous blood samples were drawn from each patient between 7 AM and 11 AM after overnight fasting. In all cases, fasting glucose levels were measured via a glucose oxidase method (Aeroset Abbott). Patients were invited to complete the International Prostate Symptom Score (IPSS), Becks Inventory for Depression (BDI), Overactive bladder (OABq) and the International Index of Erectile Function (IIEF) questionnaires at rst clinical evaluation [12][13][14]. As for our internal protocol, each man, after rst assessment, was asked to undergo semen and urine culture tests to identify potential common urogenital pathogens. Prostate speci c antigen (PSA) dosage was offered to all patients ≥ 40 yr. Realtime polymerase chain reaction (rt -PCR) platform (NIMBUS; Seegene, Seoul, South Korea) was used to detect infections by Mycoplasma, Ureaplasma, and Chlamydia spp. PCR was performed by the same laboratory for every individual. A concentration of ≥ 10 3 CFU/mL of urinary tract pathogens in the ejaculate was considered indicative of positive bacteriospermia. Likewise, a concentration ≥ 100,000 CFU/mL of urinary tract pathogens in the urine was considered indicative of a positive urine culture. The same laboratory was used for the analyses of all parameters. Data regarding history of sexually transmitted diseases (STD) (past con rmed infections of Neisseria Gonorrhea, Chlamydia Trachomatis, Herpes simplex 1 and 2 and Syphilis), history of symptomatic recurrent urinary tract infections (rUTI), and past episodes of symptomatic acute prostatitis were collected for each participant. PCa and BC diagnostic workup was performed in accordance with EAU guidelines [1]. Lastly, diagnosis of PCa and BC was achieved after expert histopathological review of prostate and bladder tissue specimens.
Data collection followed the principles outlined in the Declaration of Helsinki. All patients signed an informed consent agreeing to share their own anonymous information for future studies. The study was approved by the IRCCS San Raffaele Hospital Ethical Committee (Prot. 2014 -Pazienti Ambulatoriali).

Statistical analysis
Data are presented as medians (interquartile range; IQR) or frequencies (proportions). The analyses consisted of several statistical steps. First, patients were segregated into two groups according to the low vs. high-risk classi cation proposed by the recently revised Sexual and Reproductive Health EAU guidelines [1]. EAU high-risk patients were those with ≥ 1 of these characteristics at rst clinical assessment: i) age ≥ 40 years; ii) recurrent or persistent haemospermia (any age); iii) actual risk for PCa (e.g., positive family history); and iv) concurrent haematuria. Recurrent haemospermia was de ned as reporting ≥ 2 episodes before clinical assessment. The remaining cohort was classi ed as low risk. Second, we tested the diagnostic accuracy (sensitivity, speci city, and discrimination) of EAU guidelines.
Third, in order to improve the diagnostic accuracy of the EAU guidelines, we sought for clinical factors associated to i) positive semen culture, and ii) PCa and BC, since these factors represent the most frequent unfavourable clinical outcomes in men complaining of haemospermia. The newly identi ed factors were grouped in order to provide an updated risk strati cation. The diagnostic accuracy (sensitivity, speci city, and discrimination) of the updated risk strati cation was assessed and compared with the EAU guidelines using a DeLong test. Lastly, decision curve analysis (DCA) tested their clinical bene t. All statistical tests were two -sided with a signi cance value set at 0.05. The analyses were conducted using R (2019), a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Table 1 shows patients' characteristics of the whole cohort of 283 patients and according to the low vs. high-risk classi cation proposed by the EAU Sexual and Reproductive Health guidelines. Overall, the median (IQR) age at rst presentation was 49 (37 -48) years. Recurrent haemospermia was reported by 134 (47.4%) patients. Of all, 126 (44.5%) had a baseline CCI score ≥ 1, 67 (23.7%) had arterial hypertension, 30 (10.6%) were active smokers, 38 (13.4%) and 15 (5.3%) had a positive history for rUTI and STD at rst clinical assessment, respectively. Median (IQR) PSA at rst presentation was 1.1 (0.6 -2.4) ng/mL. Of all, 31 (11%) patients had a positive semen culture at follow -up, 16 (5.7%) a positive urine culture and 15 were found to harbour genitourinary (GU) cancers (i.e., 12 (4.2%) PCa and 3 (1.1%) BC), respectively. According to EAU guidelines, 259 (91.5%) men were classi ed as high-risk patients, whilst 24 (8.5%) low risk. High-risk patients reported higher PSA levels than low risk ones (p<0.0008). Moreover, high-risk patients were more likely to suffer of arterial hypertension (p<0.0001). On the contrary, patients in the high-risk category were similar in terms of reporting history of rUTI, history of STD, positive urine, and semen cultures after rst clinical assessment, compared to low-risk men (Table 1).     Table 3 reports the diagnostic accuracy (sensitivity, speci city, and discrimination) of the EAU guidelines compared with the updated risk strati cation. The calculated sensitivity and speci city of the EAU guidelines were 13.3% and 89.2%, respectively. The calculated sensitivity and speci city of the updated risk strati cation were 98.9% and 58%, respectively. DeLong's test con rmed that the new model [AUC  Lastly, DCA (Figure 1) displays the superior net bene t of using the new risk strati cation in terms of diagnosing positive semen cultures, PCa and BC compared to the EAU guidelines strati cation.

Discussion
We applied the EAU guidelines risk strati cation in a cohort of patients seeking rst medical help for single/recurrent episodes of haemospermia at a single tertiary academic centre. Almost nine out of ten patients were classi ed as high-risk of harbouring a treatable and identi able serious aetiology according to the most recent EAU guidelines. At rst clinical evaluation, we did not nd relevant clinical differences between high and low risk men apart from a higher prevalence of arterial hypertension and PSA levels in the high-risk group, without any difference in terms of past genitourinary tract infections. Against this background, we aimed at improving the selection of high-risk patients by identifying clinical factors associated with commonly reported "adverse outcomes" in patients with haemospermia (positive semen culture, PCa, and BC). We identi ed that patients reporting past recurrent genitourinary tract infections and with a baseline CCI ≥ 1 were the ones that should have received a focused diagnostic work-up for bacteriospermia, PCa and BC, respectively. A new risk strati cation based on these associated factors was then compared to the EAU guidelines. AUC, sensitivity, and speci city con rmed the superiority of the new risk strati cation in respect to the EAU guidelines in identifying high-risk patients ( Table 3).
As a matter of fact, our results partially con rm the ndings of previously published studies. Efesoy et al. investigated the aetiological factors among 342 patients complaining of haemospermia: the most frequent cause of haemospermia was in ammation/infection in 169 patients (49.4%), whereas genitourinary (GU) cancers were detected in only 11 patients (3.2%) [4]. Although these ndings re ect similar GU cancers prevalence to those of our cohort, we could not conclude the same regarding in ammation/infections; this difference may be due to the different and aggregate de nition/assessment of both in ammation and infection. A prospective study by Furuya et al. con rmed the overall benign aetiology of haemospermia; among the analysed cohort of 189 patients, the authors concluded that men with haemospermia without signs of infection, in ammation, or malignancy, had spontaneous resolution of their condition in more than 88% of cases [15]. Lastly, Ng YH et al. aimed to investigate the prevalence of signi cant underlying pathologies among 300 consecutive patients with haemospermia and simultaneously assess the diagnostic value of routine urological investigations. In the cohort analysed, authors found 13 PCa cases (5.7%), all in men ≥ 40 years, con rming our ndings. GU infection's rates were similar to ours, with 15% and 10.3% in men under 40 years and above 40 years, respectively [3].
Although the latter mentioned ndings are comparable to ours, it is uncertain whether these conditions are causative or coincidental with the presentation of haemospermia. Moreover, most of the published literature have been focusing in establishing the true prevalence of serious underlying diseases rather than focusing on the selection of patients that deserve adequate screening for those conditions [5,16,17].
Albeit many diagnostic tests are routinely asked in the everyday medical practice for patients with haemospermia, more testing might not always bring bene t to patient care and could lead to patient dissatisfaction along with resource misallocation. Our ndings con rm how the rigorous application of EAU guidelines may result in poor diagnostic accuracy, since virtually the entire cohort of patients should have been empirically treated and/or screened. This would have led asking for unnecessary diagnostic tests to patients that could have been treated conservatively (e.g., observation).
To our knowledge, this is the rst study to establish a new risk strati cation for the selection of high-risk patients complaining of haemospermia (single episode and/or recurrent). On the other hand, our study is certainly not devoid of limitations. First, even though we analysed a single homogeneous, same-ethnicity cohort of men presenting with haemospermia as their primary compliant, this was a single-centre crosssectional study, thus raising the possibility of selection biases. Second, although having homogeneous data from white-Caucasian patients may only represent a further strength of the analyses, on the other hand, different geographical areas and ethnicity groups might have shown different and possibly more homogenous results. Third, not all patients in our cohort underwent TRUS and/or prostate MRI thus leading to the possibility of misdiagnosing seminal vesicle stones, Mullerian duct cysts or ejaculatory duct obstructions. Fourth, our classi cation should be externally validated to con rm our ndings. On the other hand, this study suggests that the way EAU guidelines currently recommend us to stratify patients with haemospermia is not satisfactory. In the light of this, we propose here a better and more accurate risk strati cation of these men.

Conclusions
The application of the EAU risk strati cation does not adequately ensure the identi cation of high-risk patients complaining of haemospermia. In our cohort, nine out of ten patients were identi ed as high-risk according to EAU guidelines. Thus, we propose a novel and better performing risk strati cation to identify those patients at higher risk of having unfavourable associated clinical conditions.