The association of time of day and serum testosterone concentration in a large screening population

Transcription

1 Original Article TIME OF DAY AD SERUM TESTOSTEROE LEVEL I A LARGE SCREEIG POPULATIO CRAWFORD et al. Authors from the USA reviewed semen samples for their ational Prostate Cancer Awareness screening programme. They evaluated semen testosterone levels at various times during the day, depending on when the person attended for screening. In this study, they found that testosterone levels in older men remain stable throughout the morning and early afternoon, and declined only moderately after that. Further case control studies have been suggested to confirm this finding. In a study from the UK, changes to PSA screening programmes, such as lowering the threshold or incorporating reflex PSA tests, have been looked at, particularly their impact on programme performance characteristics. The authors found that such changes will increase the number of biopsies and the number of cancers detected, but the authors question whether this will improve treatment outcomes. The association of time of day and serum testosterone concentration in a large screening population E. David Crawford, Al Baha Barqawi, Colin O Donnell and Abraham Morgentaler* Surgery and Radiation Oncology, Section of Urologic Oncology, UCHSC, Aurora, CO, and *Men s Health Boston, Harvard Medical School, Boston, MA, USA Accepted for publication 9 March 27 OBJECTIVE To assess testosterone patterns during clinical hours in a large population of men participating in a national screening programme for prostate cancer, as the effect of time of day on serum testosterone concentration is unclear and largely reported in small studies. SUBJECTS AD METHODS Testosterone levels were measured in 36 men attending the national Prostate Cancer Awareness Week screening programme. Blood samples were obtained between 6. and 18. hours, whenever men presented for screening. All men completed questionnaires on age, comorbidities, height and weight. Testosterone levels were compared based on four periods, i.e. T1, (632 men), T2, (812), T3, (388), and T4, hours (1174). RESULTS The mean (SD, range) age of the men was 6.3 (9.9, 4 94) years and the mean (SD) testosterone level was (19.9) ng/dl. There was no change in mean (SD) testosterone levels over T1, T2 and T3, at (26.2), (195.8) and (181.2) ng/dl, respectively, but levels at T4, at 38.4 (176.4) ng/dl, were lower by 13% (P <.5). Advancing age, diabetes and obesity were associated with lower testosterone levels. The percentage of men with biochemical hypogonadism (<3 ng/dl) did not change across all four periods. COCLUSIOS Testosterone levels in older men are stable throughout the morning and early afternoon, declining only modestly thereafter. Further case-controlled studies are needed to confirm these findings. KEYWORDS testosterone, diurnal, ageing, diabetes, obesity ITRODUCTIO It is widely recommended that testosterone be assessed in the early morning, before 1. hours, based on studies showing a significant diurnal variation in testosterone, in which the highest levels were in the early morning with nadir levels in the evening hours [1]. However, these studies involved small samples of young men. One of the main indications for testing serum testosterone level is to aid in the diagnosis and subsequently the clinical management JOURAL COMPILATIO 27 BJU ITERATIOAL 1, doi:1.1111/j x x 59

2 CRAWFORD ET AL. TABLE 1 The overall demographics of the study population Variable Mean (SD, median) % of population Caucasian African-American Other n (%) of population (74.4) 562 (18.7) 29 (6.9) Age, years 6.3 (9.9, 6.) 61.7 (9.8) 55.8 (9.2)* 57.6 (9.2) Testosterone, ng/dl (19.9, 381.) 41.6 (187.8) (22.5)* (19.6) PSA level, ng/ml 1.7 (2.4, 1.1) BMI, kg/m (4., 26.5) 26.7 (3.9) 27.8 (4.4) 27.2 (4.) Diabetes mellitus Family history of prostate cancer 2.3 Obese (BMI 3 kg/m 2 ) 18.3 African-American 18.7 *P <.5 of male hypogonadism [2]. Late-onset hypogonadism (LOH) is common and can affect quality of life [3]. Evidence suggests that older men have substantial blunting of the diurnal variation in testosterone level [4 8]. This is not surprising, as there are substantial changes in testosterone physiology that accompany ageing. For example, testosterone levels in men decline with age, at a rate of 1%/year beginning at 4 years old [3]. Moreover, the responsiveness of the hypothalamic-pituitary axis to lower testosterone levels appears to be reduced with ageing, at least in part due to diminished pulsatility of GnRH release [9]. This leaves open the question as to the importance of time of day of testosterone testing in middle-aged and older men, particularly as this is the age group most likely to be tested for a possible diagnosis of hypogonadism. The purpose of the present study was to investigate the distribution of the diurnal variation of testosterone over the course of the working day in a large population of older men presenting for a prostate cancer screening programme. We also examined the impact of age, obesity, race and comorbidities on the relationship of testosterone levels and time of day. SUBJECTS AD METHODS Prostate Cancer Awareness Week (PCAW) is a community-based prostate cancer screening programme of the Prostate Cancer Educational Council that offers free or lowcost prostate cancer screenings at more than 4 longitudinal academic sites and single institutions across the USA. Men attending the PCAW in 23 were asked to voluntarily complete an institutional review boardapproved questionnaire that addressed questions of age, racial background, family history of prostate cancer and comorbidities. Body mass index (BMI) was calculated from self-reports of weight and height. Screening hours ranged from 6. to 18. hours, and all blood samples for PSA and total testosterone were marked by time of collection and documented in a tabulated database. Blood samples were drawn at whatever time the men presented for screening. Blood samples were collected nationwide from more that 4 longitudinal sites, and shipped frozen at or below 2 C to our laboratory. Testosterone levels were measured centrally at one laboratory using a direct chemiluminescent immunoassay (ADVIA Centaur, Bayer Healthcare AG, Germany). The minimum detectable concentration (analytical sensitivity) of this test is 1 ng/dl and the percentage crossreactivity to other compounds is <.1%, except for 5α-dihydrotestosterone at 5.4%. The precision data were obtained from manufacturer ( All consecutive men who attended the screening visit and had testosterone levels measured were included in the analysis. We excluded men who indicated a history of prostate cancer. The study population was divided into four groups based on the time of day of blood sampling, i.e. T1, (632 men), T2, (812), T3, (388), and T4, hours (1174). Hypothesis testing among the four categories of sampling time showed no significant difference between the earliest three levels of testosterone. However, the late afternoon testosterone level (after 14. hours) was significantly different from the earlier three intervals. Therefore, some data were reduced to binary coding for time of blood sampling, i.e. before 14. (early), or after 14. hours (late). Age was stratified into five categories by decade in life and testosterone level was tested as early or late within decades. The Wilcoxon rank-sum test was used to test differences in median testosterone level among the subgroups in the eighth decade in life due to the small sample size in the afternoon cohort, while a t-test was used for comparisons of other decades in life. For a given upper threshold value of testosterone, there were men who were below the threshold and presented in the morning, and men who were below the threshold but who presented in the afternoon. Similarly, there were men who presented in the morning and afternoon but who were above the threshold. Thus, at the given threshold four mean testosterone values were calculated from the four classifications of men. RESULTS The characteristics of the study population are shown in Table 1; the mean BMI was 27. kg/m 2, and 18.3% of the study population were categorized as obese, based on WHO guidelines (BMI >3 kg/m 2 ). Diabetes was present in 7.9% of the population. There were no significant diurnal changes in the median PSA levels among the four periods (P =.36). The mean testosterone levels were lower in diabetics, at 359. (168.4) ng/dl, than in non-diabetics, at 51 JOURAL COMPILATIO 27 BJU ITERATIOAL

3 TIME OF DAY AD SERUM TESTOSTEROE LEVEL I A LARGE SCREEIG POPULATIO TABLE 2 Mean age distribution by time of day Period, hours patients Mean (SD) testosterone level, ng/dl 42.1 (192.) ng/dl (P <.5), and for obese men, at (149.6) ng/dl, than in those not classed as obese, at (194.5) ng/dl (P <.5). The percentage of the study population with testosterone levels of <3 ng/dl was 31.%, and 14.1% had levels of <231 ng/dl, the threshold for hypogonadism established by several international groups [1]. Mean testosterone levels for T1 4 are shown in Fig. 1; the levels did not differ significantly age, years T1 (6. 1.) (26.2) 59.8 (9.9) T2 (1. 12.) (195.8) 6.2 (1.2) T3 ( ) (181.2) 58.3 (1.1) T4 ( ) (176.4) 61.3 (9.6) 38.8 All times (19.9) 6.3 (9.9) TABLE 3 Changes in diurnal testosterone level categorized by race Testosterone levels adjusted by age, ng/dl Mean (SD) diurnal level, ng/dl Race Total T1 3 T4 P Caucasian 41.6 (187.8) (193.5) (174.4) <.5 African-American (22.5) (26.2) (183.4) <.5 Other (19.6) 41.1 (191.) (189.8).42 FIG. 1. Mean testosterone levels with time of blood sampling ± ± ± = (23., 853.8) (23., 829.7) (29.8., 816.) (29.7, 559.9) 6 am-1 am 1 am pm FIG. 2 Mean testosterone levels early (before 14. h) and late (after 14. h) by decade of life. 7 T am 6 pm p <.1 p <.1 p <.1 p <.4 p < Decade in Life FIG. 3 Mean testosterone levels early (before 14. h) and late (after 14. h) for diabetic and obese men. T Level D D D D D O O O O Before After Diabetic (D) on-diabetic (D) Before After Obese (O) on-obese (O) between T1 and T2 (P =.28), T2 and T3 (P =.93), or T1 and T3 (P =.4), but there was a significant decline in T4 (after 14. hours) compared to all groups before 14. hours (P <.5), of 13%. Men in T4 were slightly older than men in T1 3, at 61.3 (9.6) vs 59.7 (1.1) years, respectively (P <.5); age-adjustment reduced the magnitude of the decline at T4, but this difference remained statistically significant. The mean age distribution in the four time intervals is shown in Table 2. As there was no significant difference in testosterone levels for T1 3, those groups were combined for further analyses, resulting in two groups, i.e. early and late. There was a significantly lower testosterone level in the late group for all ages studied by decade (Fig. 2). Obesity was also associated with the lower testosterone levels in the late men (P <.1), but in diabetic men there was a numerical reduction in late testosterone level that was not statistically significant (P =.12) (Fig. 3). There were small differences in testosterone levels among racial groups; African-American men had a mean testosterone level slightly higher than Caucasians, at vs 41.6 ng/ dl (P <.5), but there were no significant differences between other races (416.4 ng/dl). However, the mean age of African-American men was lower than for Caucasians (55.8 vs 61.7 years, P <.5; Table 3). The proportion of men with a testosterone level of 3 ng/dl did not change significantly during the day (P <.11) on univariate analysis. By contrast, the mean testosterone levels were lower in the late than in the early group for men with testosterone levels above this threshold (P <.5). DISCUSSIO There has been considerable recent interest in LOH, with research showing an effect of low testosterone level on a variety of systems, e.g. decreased bone mineral density [1], impaired sexuality [11], depressed mood [12], reduced muscle mass and strength [13], and altered cognitive abilities [14]. In addition, recent reports showed an association of the metabolic syndrome with low testosterone levels [15], and there is ongoing investigation into the effects of testosterone on glucose metabolism in diabetic men [13,16]. As the diagnosis of LOH requires a blood test confirming low levels of testosterone, the timing of blood sampling might be important, given the reported diurnal variation in testosterone levels. Standard recommendations have been to obtain blood samples for testosterone determination in the early morning, based on the observation that peak testosterone levels occur in the early morning, and decline rapidly after 1. hours. This recommendation to obtain early morning samples for testosterone determination was based primarily on results from studies of sequential blood sampling over 24 h in healthy young men. In middle-aged and older men the diurnal variation in testosterone levels persists, but is substantially blunted. In all of these studies, subjects were described as JOURAL COMPILATIO 27 BJU ITERATIOAL 511

4 CRAWFORD ET AL. healthy and free of endocrinopathy, and involved fewer than 2 older men [5,6,8,17]. An interesting finding by Spratt et al. [18], who measured gonadotrophins and testosterone concentrations at 2-min intervals in 2 subjects, showed an interpulse variation in LH and testosterone levels throughout the day, but the largest diurnal variation was at night. Although this study was small, this might indicate that even studies based on measuring testosterone levels in the same individual at 12-h intervals cannot be relied upon to reach conclusions for recommending a specific time to sample testosterone levels in men suspected of LOH. Further, Boyce et al. [4] reported a significant decline in both morning and evening serum testosterone level with increasing age (>4 years) and BMI in their study of 266 healthy men. In the present cross-sectional study we investigated diurnal variation in testosterone level over the course of the working day in 36 middle-aged and older ambulatory men. The major findings were that mean testosterone levels in this population were unchanged from early morning until the early afternoon (14. h), and declined only modestly thereafter, by 13%. This diurnal variation was detected for all age decades assessed. There was a decline in late afternoon testosterone levels in Caucasians and African- Americans, and for obese men (defined as a BMI of 3 kg/m 2 ). Diabetic men had a numerical decline in late afternoon testosterone levels but it was not statistically significant. To our knowledge this is the first study to investigate the potential impact of diabetes, obesity and race on diurnal variation in testosterone levels. Although there was a general pattern of a decline in testosterone later in the day almost all groups, this pattern was not universal. Specifically, there was no significant overall decline in diurnal testosterone level in diabetic men, men with testosterone levels in the hypogonadal range (<3 ng/dl), nor in men self-identified as neither Caucasian nor African-Americans, although this latter group was small. Interestingly, the percentage of diabetic men and biochemically hypogonadal men (testosterone < 3 ng/dl) did not change in each of the periods examined, particularly as there is a strong reported association between diabetes and hypogonadism [19]. This raises the possibility of an alteration in the hypothalamic-pituitary-gonadal axis in these populations. The lack of diurnal variation in hypogonadal men was reported previously by Gupta et al. [2]. In that study of morning and evening testosterone samples from the same individuals, eugonadal men had a significant decline in evening testosterone levels, but hypogonadal men did not [2]. Testosterone levels can be affected by changes in sex hormone-binding globulin and oestrogen levels that occur in obese men. This might explain the overall lower testosterone levels in obese men in the present cohort. However, the clinical significance of such association is still controversial and beyond the scope of this study. The observed loss of significance in diurnal testosterone variation between the proportion of obese men who attended the morning and the afternoon sessions can be related to the overall decrease in the total testosterone levels. This study draws strength from being large and including a diverse population, including race and health status. Although a potential study bias was the selection of a relatively highly motivated population of men who desired to be screened for prostate cancer, the demographics of this study group suggest that it is fairly representative of men seen in ambulatory healthcare facilities. This includes a substantial number of men with significant comorbidities, such as diabetes and obesity. There are several limitations to the study; the cross-sectional design cannot exclude the possibility that some men might have had a much greater degree of diurnal variation in testosterone than shown overall. There was no randomization for the time of blood collection but rather on a first-come, firstserved basis. Further, we did not exclude shift-workers from the analysis, which might have an effect on diurnal variation of testosterone in these men. Information on the use of testosteronereplacement therapy was not recorded in the study. Although the inclusion of a large cohort of men using testosterone products could influence the results of the study, this seems an unlikely confounder, as <1% of adult men in the USA are treated with testosterone products [21]. Another potential weakness of these data is the lack of corresponding bioavailable and free testosterone fractions. Although it is important to correlate variations to total testosterone levels, these data are usually not recommended clinically as a firstline screening test for LOH, and only recommended if the initial testosterone levels are suspicious, with persistent clinical symptoms suggestive of hypogonadism that might benefit from initiation of testosteronereplacement therapy [1]. onetheless, the analysis included a large cross-sectional sample which might give an overall trend representative of the diurnal variation in testosterone levels. Perhaps the greatest effect of these results relates to a practical consideration, i.e. the timing of blood testing for testosterone. As noted, the current recommendations for early morning testing for testosterone levels are not based on evidence-based data, but rather on small uncontrolled and historical studies. Given that there was no significant change in testosterone levels until 14. hours, and only a modest reduction thereafter, and the absence of a late afternoon decline in testosterone levels in men with low testosterone, it might no longer be necessary to discount testosterone levels obtained in the afternoon, or to require men seen initially in the afternoon to make an additional trip to the office or laboratory to have their blood drawn on a separate morning, to obtain an accurate test for a possible diagnosis of hypogonadism. Based on these results, we conclude that it is reasonable to liberalise the recommendations on the timing of testosterone testing for older men. In conclusion, testosterone levels in older men are stable throughout the morning and early afternoon, declining only modestly thereafter. There was no significant variation for men with testosterone levels associated with hypogonadism. COFLICT OF ITEREST one declared. REFERECES 1 Cooke RR, McIntosh JE, McIntosh RP. Circadian variation in serum free and non-shbg-bound testosterone in normal men: measurements, and simulation using a mass action model. Clin Endocrinol (Oxf) 1993; 39: JOURAL COMPILATIO 27 BJU ITERATIOAL

5 TIME OF DAY AD SERUM TESTOSTEROE LEVEL I A LARGE SCREEIG POPULATIO 2 Rhoden EL, Morgentaler A. Risks of testosterone-replacement therapy and recommendations for monitoring. Engl J Med 24; 35: Harman SM, Metter EJ, Tobin JD, Pearson J, Blackman MR. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. Baltimore Longitudinal Study of Aging. J Clin Endocrinol Metab 21; 86: Boyce MJ, Baisley KJ, Clark EV, Warrington SJ. Are published normal ranges of serum testosterone too high? Results of a cross-sectional survey of serum testosterone and luteinizing hormone in healthy men. BJU Int 24; 94: Montanini V, Simoni M, Chiossi G et al. Age-related changes in plasma dehydroepiandrosterone sulphate, cortisol, testosterone and free testosterone circadian rhythms in adult men. Horm Res 1988; 29: Bremner WJ, Vitiello MV, Prinz P. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J Clin Endocrinol Metab 1983; 56: Tenover JS, Matsumoto AM, Clifton DK, Bremner WJ. Age-related alterations in the circadian rhythms of pulsatile luteinizing hormone and testosterone secretion in healthy men. J Gerontol 1988; 43: M Diver MJ, Imtiaz KE, Ahmad AM, Vora JP, Fraser WD. Diurnal rhythms of serum total, free and bioavailable testosterone and of SHBG in middle-aged men compared with those in young men. Clin Endocrinol (Oxf) 23; 58: Veldhuis JD, Iranmanesh A, Godschalk M, Mulligan T. Older men manifest multifold synchrony disruption of reproductive neurohormone outflow. J Clin Endocrinol Metab 2; 85: ieschlag E, Swerdloff R, Behre HM et al. Investigation, treatment and monitoring of late-onset hypogonadism in males. ISA, ISSAM, and EAU recommendations. Eur Urol 25; 48: Morgentaler AA. 66-year-old man with sexual dysfunction. JAMA 24; 291: Pope HG Jr, Cohane GH, Kanayama G, Siegel AJ, Hudson JI. Testosterone gel supplementation for men with refractory depression: a randomized, placebocontrolled trial. Am J Psychiatry 23; 16: Bhasin S, Storer TW, Berman et al. Testosterone replacement increases fatfree mass and muscle size in hypogonadal men. J Clin Endocrinol Metab 1997; 82: Cherrier MM, Craft S, Matsumoto AH. Cognitive changes associated with supplementation of testosterone or dihydrotestosterone in mildly hypogonadal men: a preliminary report. J Androl 23; 24: Kupelian V, Page ST, Araujo AB, Travison TG, Bremner WJ, McKinlay JB. Low SHBG, total testosterone, and symptomatic androgen deficiency are associated with development of the metabolic syndrome in non-obese men. J Clin Endocrinol Metab 26; 91: Oh JY, Barrett-Connor E, Wedick M, Wingard DL. Endogenous sex hormones and the development of type 2 diabetes in older men and women: the Rancho Bernardo study. Diabetes Care 22; 25: Tenover JS, Bremner WJ. Circadian rhythm of serum immunoreactive inhibin in young and elderly men. J Gerontol 1991; 46: M Spratt DI, O Dea LS, Schoenfeld D, Butler J, Rao P, Crowley WF Jr. euroendocrine-gonadal axis in men: frequent sampling of LH, FSH, and testosterone. Am J Physiol 1988; 254: E Corona G, Mannucci E, Petrone L et al. Association of hypogonadism and type II diabetes in men attending an outpatient erectile dysfunction clinic. Int J Impot Res 26; 18: Gupta SK, Lindemulder EA, Sathyan G. Modeling of circadian testosterone in healthy men and hypogonadal men. J Clin Pharmacol 2; 4: Institute of Medicine of The ational Academies. Testosterone and Aging. Clinical Research Directions. Liverman CT ed Committee on Assessing the eed for Clinical Trials of Testosterone Replacement Therapy. Washington, DC: The ational Academies Press, 24: Correspondence: Al Baha Barqawi, Assistant Professor in Surgery/Urology, Director of Research/Section Urologic Oncology, Anschutz Cancer Pavilion, Ursula St. Suite 14, Mail Stop: F-71, Aurora, CO 81 51, USA. al.barqawi@uchsc.edu Abbreviations: LOH, late-onset hypogonadism; BMI, body mass index; PCAW, Prostate Cancer Awareness Week. JOURAL COMPILATIO 27 BJU ITERATIOAL 513