Severe erectile dysfunction is a marker for hyperprolactinemia

(2001) 13, 176±182 ß 2001 Nature Publishing Group All rights reserved 0955-9930/01 $15.00 www.nature.com/ijir Severe erectile dysfunction is a marker for hyperprolactinemia AM Johri 1, JPW Heaton 1 * and A Morales 1 1 Queen's University, Kingston General Hospital, Ontario, Canada The need for routine prolactin (PRL) measurement in the initial evaluation of erectile dysfunction (ED) has been questioned because of the low rate of hyperprolactinemia (HP) in these men and the costs involved. In addition, it is widely thought that sexual desire problems are a good clinical marker for HP and=or low testosterone in men with ED. Within a 15-month period, 844 consecutive PRL and sexual hormone determinations were conducted in men at the Kingston General Hospital. Of these patients, 138 were comprehensively evaluated at the rst visit for ED and completed the International Index of Erectile Function (IIEF). In the 138 patients, 2.2% had severe hyperprolactinemia (>35 ng=ml), within the range of 1 ± 5% previously reported. No correlation between initial prolactin value and the sexual desire domain or the erectile function domain (EFD) of the IIEF was found for this population. However, all cases of severe HP were found to occur in men who scored less than 10 in the EFD of the IIEF. Low libido is widely accepted as a marker of HP. In this study, HP was found in patients not reporting major problems with a desire disorder. Clinically signi cant HP may be reliably found with routine biochemical evaluation and in this series was not detected in patients with EFD scores above 10. A routine PRL measurement is inexpensive and early detection of a serious and treatable disease may afford greater therapeutic success. (2001) 13, 176±182. Keywords: prolactin; erectile dysfunction; libido; routine measurement Introduction The need for routine prolactin measurement in the initial evaluation of erectile dysfunction (ED) has been questioned because of the low rate of hyperprolactinemia (HP) in these men and the cost of conducting routine measurements. 1,2 Furthermore, it is commonly accepted that poor libido is an adequate clinical marker for HP and=or low testosterone in men with ED and so may serve as the basis for selective prolactin (PRL) determinations. 3,10 This is based on the observation that decreased libido levels may be correlated to increases in that individual's PRL level. However can a single initial assessment of libido be predictive of HP for any given man presenting with ED? If so, then the initial libido status can be used to conduct selective PRL measurements. To address this question, we have compared the initial libido status (using various indices of sexual desire) with corresponding initial PRL levels of men with ED. To address the cost of routine PRL measurements, we have determined the prevalence of HP and calculated the cost of detecting a single case in the ED clinic. *Correspondence: JPW Heaton, MD, Queen's University, Kingston General Hospital, 76 Stuart Street, Kingston, Ontario, Canada, K7L 2V7. Received 1 August 2000; accepted 26 February 2001 The increased availability of effective treatments for ED which do not address the underlying disease presents a challenge to the accurate diagnosis of ED. 4 There are now reports indicating that successful treatment of ED may occur despite potentially signi cant HP. 5 A case of missed HP in which sildena l improved sexual function has been described. 5 To demonstrate the value of routine PRL measurement in all men presenting to the ED clinic, this report presents an additional case of a man whose ED initially improved with sildena l. The result from the routine PRL measurement was investigated at a later time and indicated that the underlying cause of his ED was in fact HP. Materials and methods Within a 15-month period (4 January 1999 ± 13 April 2000), 844 consecutive PRL and sexual hormone determinations were conducted in men at the Kingston General and af liated Queen's University hospitals. Three hundred and seventy-four of these men were examined by a physician for a urological condition. Of these men, 138 were comprehensively evaluated at the rst visit for ED and completed the International Index of Erectile Function (IIEF). 6 A thorough sexual, medical and psychosocial history

was taken and a focused physical examination was performed on every patient with ED. 7 Eighty-seven patients also answered a single question asked by an experienced interviewer in a customized inventory (CI) of sexual status designed to ascertain the libido level (increased, stable, or decreased). ED was de ned as dif culty in initiating and=or maintaining an adequate erection to effect penetration and ejaculation. 8 Erectile function and sexual desire were determined by the Erectile Function Domain (EFD) and Sexual Desire Domain (SDD) of the IIEF. All men presenting to the ED clinic had a routine PRL measurement conducted at the rst visit. All serum values were obtained by the Kingston General Hospital Biochemistry Laboratory. Free and total testosterone, luteinizing hormone (LH), follicular stimulating hormone (FSH) and glucose were measured using standard techniques. PRL was measured using the Bayer Immuno 1 Automated Immunoassay Instrument (Tarrytown, NY). The total number of men with high PRL levels requiring treatment were reported. Characteristics of all found cases were evaluated. One instructive case of missed HP, initially treated with sildena l is described in full. The total cost of conducting a routine PRL measurement, and the cost of detecting a single case of true HP in the 138 patients was calculated. We also analysed data from a new group of patients presenting to the ED clinic from January 1997 to January 1999, in order to con rm our initial ndings. Results The average PRL levels per decade for 427 determinations (374 patients) ordered in association with a urological condition was determined as detailed in Table 1. Complete PRL and IIEF scores were available for 138 cases. Three patients out of these 138 (2.2%) had PRL levels above 35 ng=ml which corresponds to markedly high or severe hyperprolactinemia. 2 Six patients out of 138 had mild HP de ned as being 15 ± 35 ng=ml. 2 A statistical correlation between initial testosterone and PRL levels was not found for the group of men with ED (Figure 1). However, all cases of severe HP were found to occur below 10 nmol=l. Many hypogonadals were found not to have HP. Four out of six cases of mild HP were found to occur within the normal range of testosterone. There was no statistical correlation between the initial SDD score and testosterone levels for this population of men (Figure 2A). Although PRL is commonly associated with low desire in individuals, 2,10 the initial SDD scores for this population were not found to correlate with the corresponding PRL levels obtained (Figure 2B). The three cases of severe HP found were associated with a SDD score of less than 5. Testosterone levels were not statistically correlated with the EFD score (Figure 2C). A surprising relationship between PRL and erectile function was found. All men with mild and severe HP were found to have an EFD score below 10, suggesting a clinically important effect of PRL on erectile function (Figure 2D). To con rm these ndings, a new group of patients was analysed. Analysis of PRL levels determined between 1997 and 4 January 1999 revealed that all cases of HP were again found to occur in men who scored less than 10 in the EFD of the IIEF (Figure 2D). This study revealed no statistical correlation between the initial SDD score of the IIEF and PRL, but can a correlation be found using an alternate index of desire? Data generated from the single question (`Is your libido decreased?') in a CI of sexual status was available for 87 patients. The average PRL levels in patients reporting decreased libido in the CI was not signi cantly different from those reporting normal libido (Figure 3A). This suggests that the CI, and again the initial desire 177 Table 1 Mean and median prolactin distribution by decade for patients assessed for a urological condition in a 15-month period Patient age Total no patients (no of prolactin determinations)* Prolactin ng=ml Mean (ranges) Median 21 ± 30 5 (5) 5.1 (3.3 ± 6.7) 4.6 31 ± 40 20 (23) 8.8 (2.8 ± 20.3) 7.5 41 ± 50 63 (72) 10.0 (1.7 ± 164.4) 4.5 51 ± 60 117 (130) 5.2 (1.9 ± 23.9) 4.8 61 ± 70 108 (125) 13.2 (1.2 ± 366) 5.4 71 ± 80 55 (66) 102.3 (0.1 ± 4235) 5.3 81 ± 90 6 (6) 8.6 (5.2 ± 16.1) 6.4 Totals 374 (427) *Some patients had repeat prolactin determinations. The means are obtained from all prolactin determinations. Figure 1 Scatter plot of testosterone values and corresponding prolactin levels measured in 138 men presenting to the Erectile Dysfunction Clinic.

178 Figure 2 (A) Scatter plot of testosterone levels with the corresponding score from the sexual desire domain of the IIEF (January 1999 to April 2000, n ˆ 138). (B) Scatter plot of prolactin levels with corresponding scores from the sexual desire domain of the IIEF (January 1999 to April 2000, n ˆ 138). (C) Scatter plot of testosterone levels with corresponding score from the erectile function domain of the IIEF (January 1999 to April 2000, n ˆ 138). (D) Scatter plot of prolactin levels with the corresponding score from the erectile function domain of the IIEF. status, were not predictive of PRL levels in this study. As expected, the average SDD score of patients reporting decreased libido in response to the CI was lower than the average score of patients reporting normal libido (Figure 3B). The average EFD score from the IIEF of patients reporting decreased libido in the CI was not signi cantly different from those patients reporting normal libido in the CI (Figure 3C). The average testosterone level in patients reporting decreased libido was not signi cantly different from those reporting normal libido (Figure 3D). In summary, the psychometric indices of desire and erectile function used routinely by our clinic were not predictive of HP or hypogonadism. However, all cases of HP occurred in patients with an EFD score below 10. Case study The patient was a 50-y-old white man with no signi cant past medical history who was referred to the ED clinic. The patient presented with poor sexual function in a new relationship with a signi cant anxiety interaction component. The IIEF scores were rated as follows: Erectile Function Domain (6=30), Orgasmic Function Domain (2=10), Sexual Desire Domain (4=10), and Intercourse

Satisfaction Domain (4=15). The man reported normal libido in response to the CI of sexual status. General chemistry revealed low total testosterone (3.0 nmol=l) and high PRL (164.4 mg=l) levels. Repeat determination con rmed these values. PSA and DHEAS were within normal limits. A trial of sildena l treatment (Viagra TM, P zer, NY) was prescribed. Sildena l treatment resulted in an immediate improvement in erectile function. Once the PRL level had been con rmed, a CT scan was conducted. The CT could not rule out a microadenoma and so the patient's therapy was switched 179 Figure 3 Libido status as determined by the single question (`Is your libido decreased or normal?') in a customized inventory of sexual status compared with clinical indices. (A) Average prolactin level and libido (n ˆ 87). (B) Average Sexual Desire Domain score and libido (n ˆ 87). (C) Average Erectile Function Domain score and libido (n ˆ 87). (D) Average testosterone and libido (n ˆ 87).

180 to bromocriptine. Within 2 weeks of bromocriptine treatment (2.5 mg), the patient's sexual function was improved in all respects as identi ed by patient and partner. After 4 weeks of bromocriptine therapy the IIEF score was rated as 30=30 for the EFD without any use of sildena l. PRL levels were stabilized within the normal range. The patient elected to remain on bromocriptine therapy. Discussion Consensus on the role of routine PRL measurement in the initial evaluation of men with ED has not been established. Akpunonu et al evaluated the prevalence of HP in 212 impotent men and found the prevalence too low (1.4%) to justify a routine PRL screen at their institutional charge of $57 for each determination. 1 Selective determination in patients with low testosterone was suggested. We were unable to t a curve to the initial testosterone levels and corresponding PRL levels for a population of men with ED. Testosterone may fall as PRL rises in an individual, but for our population of men, the initial testosterone level was not predictive of the PRL level. As seen in Figure 1A, several men with low initial testosterone levels had normal or low PRL levels. Determination of PRL only in the case of markedly low initial testosterone (below 4 nmol=l) would have missed ve of the six cases of mild HP and two of the three cases of severe HP. Determination in cases of testosterone below 10 nmol=l would have encompassed all cases of severe but not mild HP. Johnson et al found a prevalence of 2.1% for HP in 330 impotent patients and recommended screening men only with low libido or bilateral testicular atrophy considering that their cost for a PRL level was $92. 9 We found a similar prevalence of severe HP (2.2%) but no clear association between libido and PRL at the population level. HP has been associated with loss of libido in individuals, 10 but can an initial assessment of libido on its own predict whether an individual will have de ned HP? We found that if determining PRL only in cases of low sexual desire (SDD score below 3), more than 50% of the cases of HP would have been missed. Measuring PRL when the SDD was 5 or below would encompass the cases of severe but not mild HP. When applied to a population of men with ED, the SDD of the IIEF and the direct questions about sexual desire and libido (the CI) do not select for patients likely to have high PRL. Within an individual, the changes in the SDD or CI may be so striking when the HP is treated that there is no doubt about the link, for that individual, between PRL and libido. However these instruments appear not to be precise enough, and PRL, libido, and testosterone are subject to so many unrelated in uences, such that the instruments have no discriminatory value in terms of deciding when to measure serum PRL at rst contact. In keeping with the distinction made between an individual and a population of men with ED, we were unable to nd a statistical correlation between the initial EFD score of the IIEF and PRL levels for the group studied. However, 100% of the cases of mild to severe HP were found to occur in men who scored less than 10 in the EFD of the IIEF questionnaire. This score has been shown to agree with a severe degree of ED. 11 Though a low EFD score is not predictive of HP (many more men with low scores had normal to low PRL, Figure 2D), this value serves as a useful cut-off point, below which, one cannot elect to forgo PRL determination. When evaluating the association between HP and disease, it must be considered that several PRL variants have been detected both in normal individuals and in a variety of disease states. 12 The high molecular weight forms which are named `big' (50 ± 60 kda) and `big ± big' (150 kda) PRL are less active in vivo and=or are transported less easily across the capillary bed than the 22 kda molecules. 13 `Big' and `big ± big' PRL (macroprolactinemia) do not appear to contribute to ED. 14,15 For this reason, Cavaco et al. suggest PRL measuring methods which are less sensitive to the presence of `big ± big' PRL in order to differentiate clinically important HP from that caused by macroprolactinemia. 16 The assay used to measure PRL in this study could not exclude macroprolactinemia and was unable to identify which PRL variant was being measured. It is therefore possible that some cases of HP detected in this study could be due to the presence of `big' or `big ± big' PRL. In these cases, the ED may be associated with psychological ED rather than being associated with PRL. A PRL assay insensitive to `big' or `big ± big' PRL would be the ideal method to avoid these issues. The rate of HP obtained in this study (2.2%) ts within the 1 ± 5% range of previous studies. 1,2,9,17 ± 19 Compilation of these studies with ours results in prevalences of PRL greater than 35 ng=ml at 0.9% (28 out of 3265) and of pituitary adenomas at 0.4% (18 out of 4501). If a PRL determination had not been conducted in the 138 cases studied, nine cases of HP above 15 ng=ml, and three cases above 35 ng=ml could have been missed. The cost of conducting a routine PRL measurement in the 138 men was $182 at the incremental cost of $1.32 per determination at this institution (all gures $US). The cost of detecting a single case of severe HP is only $60.72 at this institution. Utilizing a private laboratory in our region, the cost of detecting a case of severe HP in men presenting to the ED clinic would be $851 at a charge of $18.50 per PRL determination. Although it is dif cult to compare costs between institutions with variable cost contributors and billing systems, it is clear that in this environment, a routine PRL

level is a relatively inexpensive method of detecting a serious but reversible disease process. Previous studies identi ed PRL levels above 35 ng=ml as markedly high. 2 Our analysis of 138 impotent men showed that two men out of six with PRL levels between 15 and 35 ng=ml were found to have a pituitary microadenoma. Therefore further investigation of the etiology of ED in men with even moderate HP may aid in the initial diagnosis. Finally, this report suggests that with the advent of medication which improves ED without addressing the underlying disorder, a routine PRL measurement in all men presenting to the ED clinic will avoid missing HP. Our case study describes a man who was initially treated with sildena l for ED. Despite the improvement in sexual function with sildena l, the result of a routine PRL measurement was fully investigated at a later time to comply with the routine protocol used in this clinic for the initial assessment of ED. The existence of this protocol was fortunate for the patient as the routine PRL measurement revealed HP. The patient's therapy was then switched to bromocriptine. Thus timely detection of HP allowed treatment of the underlying cause of sexual dysfunction. HP resulting from pituitary adenomas can have serious consequences if left unrecognized. 20 Individuals with such lesions may present with loss of libido, ED, ejaculatory incompetence, reduced seminal volume, gynecomastia, and visual eld disturbances and local compressive symptoms in males. 21 As reported previously, in a series of 22 male patients with PRL-secreting pituitary tumors, 91% had ED and decreased libido. 22 The man in this case study had no clinical signs of HP other than ED, and reported normal libido in response to at least one index of desire used. The absence of clinical markers of HP in such cases illustrates the importance of a full assessment for the cause of ED before any therapy in order to avoid missing the diagnosis of a potentially curable or serious disease. 23 Conclusion Our analysis of 138 men provides systematic evidence to support the `Process of Care' model recommendation that a routine PRL level determination should be conducted in the initial evaluation of ED. 7 Data generated by psychometric indices of desire (such as the SDD of the IIEF and the single question CI) did not serve as a predictor for both mild and severe HP. Low testosterone alone cannot substitute as a predictor for HP. Selective PRL determination based on these indices would result in missed cases of HP. If PRL can only be determined by an institution in select cases, then a low EFD score on the IIEF (below 10) should be included as an additional ag for testing PRL levels. Our data suggests that the safest approach would be to conduct routine PRL measurements in all men presenting to the ED clinic. This becomes especially important in light of the increased use of medications that improve ED (and some that may affect libido) without addressing the underlying pathology. A routine PRL measurement is relatively inexpensive in this environment and early detection of a potentially serious and treatable disease may afford greater therapeutic success. Acknowledgements The assistance of Dr Michael J Raymond (Chief, Clinical Chemistry), and the nancial support from the Queen's University Dean of Medicine, is gratefully acknowledged. References 1 Akpunonu BE et al. Routine prolactin measurement is not necessary in the initial evaluation of male impotence. J Gen Intern Med 1994; 9: 336 ± 338. 2 Buvat J, Lemaire A. Endocrine screening in 1022 men with erectile dysfunction: clinical signi cance and cost-effective strategy. J Urol 1997; 158: 1764 ± 1767. 3 Nickel JC et al. Endocrine dysfunction in impotence: incidence, signi cance, and cost-effective screening. J Urol 1984; 132: 40 ± 43. 4 Seftel AD. Editorial: erectile dysfunction Ð the paradigm changes but the challenges remain. J Urol 1997; 156: 1768 ± 1769. 5 Garg RK, Khaishgi A, Dandona P. Is management with sildena l changing clinical practice? Lancet 1999; 353: 375 ± 376. 6 Rosen RC et al. The International Index of Erectile Function (IIEF): a multidimensional scale for assessment of erectile dyfunction. Urology 1997; 49: 822 ± 830. 7 Padma-Nathan H. Diagnostic and treatment strategies for erectile dysfunction: the `Process of Care' model. Int J Impot Res 2000; 12(Suppl 4): S119 ± S121. 8 NIH Consensus Development Panel on Impotence. Impotence. JAMA 1993; 270: 83 ± 90. 9 Johnson AR, Jarow JP. Is routine endocrine testing of impotent men necessary? J Urol 1992; 147: 1542 ± 1544. 10 Foster RS et al. Role of serum prolactin determination in evaluation of impotent patient. Urology 1990; XXXV1: 499 ± 501. 11 Cappelleri JC et al. Diagnostic evaluation of the erectile function domain of the International Index of Erectile Function. Urology 1999; 54: 346 ± 349. 12 Sinha YN. Structural variants of prolactin: occurrence and physiological signi cance. Endocrine Rev 1995; 16: 354 ± 369. 13 Leite V et al. Characterization of big ± big prolactin in patients with hyperprolactinemia. Clin Endocrinol 1992; 37: 365 ± 372. 14 Guay AT, Sabharwal P, Varma S, Malarkey WB. Delayed diagnosis of psychological erectile dysfunction because of the presence of macroprolactinemia. J Clin Endocrinol Metab 1996; 81: 2512 ± 2514. 15 Tritos NA, Guay AT, Malarkey WB. Asymptomatic `big' hyperprolactinemia in two men with pituitary adenomas. Eur J Endocrinol 1998; 138: 82 ± 85. 181

182 16 Cavaco B et al. Hyperprolactinemia due to big ± big prolactin is differently detected by commercially available immunoassays. J Endocrinol Invest 1999; 22: 203 ± 208. 17 Kropman RF et al. Routine endocrine screening in impotence: signi cance and cost-effectiveness. Int J Impot Res 1991; 3: 87. 18 Maatman TJ, Montague DK. Routine endocrine screening in impotence. Urology 1986; 27: 499. 19 Leonard MP, Nickel CJ, Morales A. Hyperprolactinemia and impotence: why, when, and how to investigate. J Urol 1989; 142: 992. 20 Zonszein J. Diagnosis and management of endocrine disorders of erectile dysfunction. Urol Clin N Am 1995; 22: 789 ± 802. 21 Besser GM et al. Clinical and therapeutic aspects of hyperprolactinemia. In: MacLeod RM, Thorner MO, Scapagnini U (eds). Prolactin. Basic and Clinical Correlates. Liviana Press: Padova, 1985, pp 833 ± 847. 22 Carter JN et al. Prolactin-secreting tumors and hypogonadism in 22 men. New Engl J Med 1978; 299: 847 ± 852. 23 Chan-Tack KM. Oral sildena l in erectile dysfunction. Lancet 1998; 352: 1557.