Pituitary Evaluation in Patients With Low Prostate-Specific Antigen

Andjela Drincic, MD; Anh-Thu Nguyen, MD; Shilpi Singh, MBBS; Mohsen Zena, MD; Ryan Walters, PhD; Kathryn Friedman, RN; Robert J. Anderson, MD


Endocr Pract. 2018;24(12):1030-1037. 

In This Article


There are very limited data that address the prevalence of low level of PSA in a general male population. Most papers that reported the PSA reference ranges focused on defining the upper normal limit and typically report the range of PSA values from 25th to 75th percentile as well as the 95th percentile. As a result, while the actual PSA minimal value range is measured, the findings are not readily available. Based on data from Oesterling et al,[2] it appears that a PSA of 0.1 ng/dL accounts for <2.5th percentiles for serum PSA values of 471 men, aged 40 to 80 years.

In our low-PSA cohort, all subjects as a group had total testosterone values in the lowest quartile for the reference range. The prevalence of hypogonadism in the low-PSA group was significantly higher than in the control group (45.5% vs. 15.0%).

We further evaluated both of our groups for conditions known to contribute to hypogonadism, such as age, T2DM, obesity, and narcotic therapy. All subjects in our study were older (mean age >60 years), but as shown in Table 1, there was no difference in age between the low-and normal-PSA groups. While total testosterone levels decline with age,[15,16] in general, this decrease is small at about 1% per year.[17] Estimated prevalence of late-onset hypogonadism at 5 to 25% depends on the diagnostic criteria used,[18–22] with weight gain and deterioration of general health as the major contributing factors.[23] In our study, the prevalence of hypogonadism in the control group at 15% was within the expected rate, while the prevalence of hypogonadism at 45.5% in our low-PSA cohort was well above expectation for the general population. Prevalence of T2DM in both our groups was similar but overall higher than in the general population, at 43.2% in the low-PSA group and 30% in the control group (P = .411). This is relevant, because about 25 to 40% of men with T2DM have hypogonadotropic hypogonadism.[24–27] Men with T2DM have 20% lower PSA concentrations than those without.[28] Hypogonadal men with T2DM have a lower PSA than their eugonadal counterparts.[29] Obesity, another factor associated with a high rate of hypogonadism,[30] was also prevalent in our subjects, but there was no BMI difference between the two groups. The average BMI in our cohort and control group was 33.9 ± 8.2 kg/m2 and 31.8 ± 5.2 kg/m2 (P = .201), respectively. Serum testosterone in a man with BMI >30 kg/m2 is approximately 30% lower than that of a man with BMI <25 kg/m2 at any age.[31] Finally, the association of narcotic use and hypogonadism is well recognized.[14,32] Again, there was a high but similar use of narcotics in both groups: 25% in the low-PSA group and 20% in the control group (P = .201).

The prevalence of a low IGF-1 value in our low-PSA cohort, which is strongly suggestive of GH insufficiency,[33] was impressive, at 18.6%. This is likely partly responsible for the decreased prostate volume which correlates with low PSA.

In the low-PSA group, 5 of 44 subjects (11.4%) were diagnosed with either partial or complete hypopituitarism. This dramatically exceeds the prevalence of hypopituitarism in the general population. In a European study that evaluated 146,000 subjects, based on baseline and hormonal dynamic tests, the prevalence of hypopituitarism was estimated to be between 29 and 45.5 per 100,000.[34] All our men with hypopituitarism were asymptomatic, which is not surprising, as hypopituitarism in men is often clinically silent. Importantly, in 3 out of 4 subjects with clinically indicated pituitary MRI, the imaging showed abnormalities.

The strength of this study is that it is the first clinical study to evaluate the pituitary hormonal profile in asymptomatic men found to have a low PSA. This is important, as the PSA is still a very frequently ordered laboratory test, and patients with low to undetectable PSA are not rare. In our cohort, the prevalence of hormonal abnormalities, from hypogonadism to low IGF-1 to hypopituitarism, was notable. Our results support the need for hormonal evaluation of asymptomatic subjects with a very low PSA. Another important strength is that we recruited a control group made up of subjects who were age- and BMI-matched and had similar prevalence of major factors known to be associated with hypogonadism, as outlined above. All subjects had full hormonal evaluation utilizing the same assays.

The limitations of our study are related to difficulties in the interpretation of testosterone values we obtained and the comparisons to the normative data reported in the literature due to laboratory methodologic issues. It is well recognized that currently recognized reference intervals for testosterone vary widely depending on the specific assay and population used to establish the intervals, complicating the interpretation of the results.[35,36] While total plasma testosterone is most reliably measured by liquid chromatography–tandem mass spectrometry, it is generally considered that radioimmunoassays and immunometric assays, as available in most hospital laboratories, are sufficiently accurate to distinguish eugonadal from hypogonadal men.[14,35] Another limitation is related to the definition of the threshold for the lower normal range for total testosterone.[36] This level was most recently reported to be <275 ng/dL,[37] but our laboratory at the VA-NWIHCS reported 170 ng/mL for the veteran population. In addition, we based our diagnosis of hypogonadism on a single morning testosterone level (usually, two measurements are recommended to confirm the hypogonadism). However, given that the exclusion criteria were robust, none of the subjects had any acute illness, and the testosterone threshold to diagnose hypogonadism was fairly low (<170 ng/dL as opposed to the usual <275 ng/dL for the lower limit of testosterone), we believe that we captured an accurate prevalence of hypogonadism in this pilot study.

Another limitation of this study is lack of reliable free testosterone data for the study population. This is significant given the fact that a major proportion of these men in the low-PSA group (43%) had diabetes and a majority were obese, with a mean BMI of 33.9 kg/m2. Type 2 diabetes and obesity are associated with low sex hormone–binding globulin, which can give a false low value of total testosterone. In these situations, it is helpful to measure sex hormone–binding globulin to calculate the free testosterone or measure free testosterone by equilibrium dialysis; however, those were not available in our institution at the time. Free testosterone in the VA Health Care System was measured with the Siemens Coat-A-Count free testosterone solid-phase [125]I radioimmunoassay. Twenty-eight subjects (63.6%) had a low free testosterone using such measurement. However, sex hormone–binding globulin data were not available for all patients, and the assay used was a less optimal analogue assay. Thus, we cannot report free testosterone data that are reliable enough to draw definite conclusions.

We did not design our study to detect the true prevalence of GH insufficiency in the cohorts evaluated. While low IGF-1 in the absence of liver disease is strongly suggestive of GH deficiency, a normal IGF-1 does not exclude GH deficiency.[33] Additional tests, including the insulin tolerance test, would have been needed to detect GH deficiency more reliably in these subjects.

While the presence of an age- and BMI-matched control group is a notable strength of our study, we should acknowledge that our control group had only 20 patients (compared to 44 in the low-PSA group) due to feasibility and cost issues. It is helpful to know that there was no difference between the groups in prevalence of diabetes, OSA, or use of narcotics and that the distribution of weight was grossly similar. Still, the difference in group sizes could have potentially influenced the results of our hormonal findings. This indeed is a pilot study, and future research is needed to confirm our results.

Finally, we are unable to comment on the prevalence of a very low PSA in men seeking care at the VA-NWIHCS because of technical limitations. The computer-generated list of low PSA was inaccurate in distinguishing those with prostate disease from those without it. Manual accurate extraction of data was not physically possible for such a large group of patients. Many patients who seemingly met the criteria chose not to enroll in the study.

In summary, a low PSA value was associated with hypogonadism in our cohort. While the majority of men had chronic conditions predisposing them to hypogonadism, some had other important pathologies, such as pituitary tumors. Men with a low or undetectable PSA need to be further clinically evaluated for signs and symptoms of hypogonadism. Our data indicate that a PSA ≤0.1 ng/mL may be another indication for measurement of testosterone level.