Origin of the concern regarding testosterone and prostate cancer

In 1941 Charles Huggins and Clarence Hodges published the first report indicating that castration or treatment with estrogen in men with metastatic PCa caused a decline in the serum marker acid phosphatase. Both treatments had the known effect of lowering serum T, and it was concluded that T caused an “enhanced rate of growth” of PCa. Androgen deprivation has been a mainstay of treatment for advanced PCa since that time. In addition, the investigators reported that T administration in these same men caused immediate PCa “activation” in all the individuals who received it.

In 1981 Fowler and Whitmore reviewed the experience with T administration in men with metastatic PCa at Memorial Sloan Kettering Cancer Center. Of 52 men who received T, 45 had an “unfavorable” response, most within 30 days. This study led to the widely repeated comment that offering T to a man with PCa was like “pouring gasoline on a fire,” or “feeding a hungry tumor.”

However, reexamination of these historical articles revealed something quite different. In the article by Huggins and Hodges, 3 men were reported to receive T, yet results were only provided for 2 of these men, and 1 of these 2 had already been castrated. Because we know today that an increase in serum T after androgen deprivation causes an increase in prostate markers such as PSA even in men without cancer, we can disqualify this individual from contributing to any general rule about T and PCa. In the end, the landmark article by Huggins and Hodges provides information on only a single man who received T administration without prior androgen deprivation. In this single case, the acid phosphatase results were erratic.

The study by Fowler and Whitmore is even more instructive. Of the 52 men, only 4 were not yet androgen deprived by either castration or estrogen administration. Of these 4 men, 1 had an early unfavorable result, and the other 3 continued to receive daily injections of T for 52, 55, and 310 days, without apparent negative effects. The investigators themselves were so surprised by these results that they postulated that naturally occurring serum T concentrations may be sufficient to produce maximal growth of PCa.

What a reexamination of these and other articles from the pre-PSA era tell us is that T administration in men who were previously androgen deprived resulted in a rapid and near-universal growth in PCa manifestations, whereas T administration in hormonally intact men produced little or no effect. This finding suggests that there is a limit to the ability of androgens to stimulate PCa growth.

Longitudinal studies

The relationship of T and other sex hormones to subsequent development of PCa has been extensively studied, in at least 21 population-based longitudinal studies. In these studies a health history is obtained, and blood samples at baseline are then frozen for the duration of the study, in some cases up to 20 years or longer. At the end of the study men who have developed PCa are identified, and a matched set of men without PCa serve as controls.

Most of these studies surprisingly have shown no influence of serum T or other androgens on PCa risk. A small number have revealed weak associations with minor androgens, or ratios of one hormone to another, and in one case there was an association between quartile levels of free testosterone and PCa risk. However, these results have not been reproducible. It is worth noting that one large study of this type actually noted reduced PCa risk in men with higher T levels.

In 2008 the investigators of 18 of these longitudinal studies pooled their data to form one large data set, creating enough power to examine a variety of questions regarding serum sex hormones and PCa. In contrast to a meta-analysis, this was a single, actual study. With more than 3886 men with PCa and more than 6438 men as age-matched controls, this was one of the largest PCa studies published to date. The results showed no association between any androgens and PCa risk. Men with higher endogenous concentrations of serum T were at no greater risk of developing PCa than men with low levels of serum T.

The importance of these studies is that they provide a sophisticated method of investigation to determine the long-term effects of endogenous hormone levels, especially testosterone, on the subsequent risk of development of PCa. Although such studies cannot entirely replace the value of a prospective long-term controlled study of TTh, they do address the question as to whether high levels of T (or other hormones) predispose men to a greater risk of later development of PCa. On this question, these prospective longitudinal studies provide two uniform and convincing answers: first, that men who develop PCa do not have higher baseline T levels, and second, men with higher T levels are at no greater risk of developing PCa than men with lower T concentrations.

Longitudinal studies

The relationship of T and other sex hormones to subsequent development of PCa has been extensively studied, in at least 21 population-based longitudinal studies. In these studies a health history is obtained, and blood samples at baseline are then frozen for the duration of the study, in some cases up to 20 years or longer. At the end of the study men who have developed PCa are identified, and a matched set of men without PCa serve as controls.

Most of these studies surprisingly have shown no influence of serum T or other androgens on PCa risk. A small number have revealed weak associations with minor androgens, or ratios of one hormone to another, and in one case there was an association between quartile levels of free testosterone and PCa risk. However, these results have not been reproducible. It is worth noting that one large study of this type actually noted reduced PCa risk in men with higher T levels.

In 2008 the investigators of 18 of these longitudinal studies pooled their data to form one large data set, creating enough power to examine a variety of questions regarding serum sex hormones and PCa. In contrast to a meta-analysis, this was a single, actual study. With more than 3886 men with PCa and more than 6438 men as age-matched controls, this was one of the largest PCa studies published to date. The results showed no association between any androgens and PCa risk. Men with higher endogenous concentrations of serum T were at no greater risk of developing PCa than men with low levels of serum T.

The importance of these studies is that they provide a sophisticated method of investigation to determine the long-term effects of endogenous hormone levels, especially testosterone, on the subsequent risk of development of PCa. Although such studies cannot entirely replace the value of a prospective long-term controlled study of TTh, they do address the question as to whether high levels of T (or other hormones) predispose men to a greater risk of later development of PCa. On this question, these prospective longitudinal studies provide two uniform and convincing answers: first, that men who develop PCa do not have higher baseline T levels, and second, men with higher T levels are at no greater risk of developing PCa than men with lower T concentrations.

Testosterone trials

In the absence of any single large study on TTh, one must examine the results from smaller studies. One of these was a 12-month study of 371 men on T gel therapy. Over the course of 1 year 3 cancers were detected, all due to an increase in PSA. One of these increases in PSA was transient and resolved; however, a biopsy was performed and revealed cancer. In this study the mean increase in PSA was 0.4 ng/mL. This increase was noted at 3 months, and PSA remained unchanged over the next 9 months.

Other studies have revealed a similar rate of cancer detection in TTh trials. In a review of 9 separate TTh trials involving 579 men and ranging from 3 to 36 months, 7 cancers were identified, representing a cancer detection rate of 1.2%. Wang and colleagues performed one of the longest TTh trials. In this study 163 men with a mean age of 51 years received T gel for 42 months. Over this time the mean PSA increased from 0.85 ng/mL at baseline to 1.1 ng/mL at 6 months, and then did not change significantly over the next 3 years of the study. Three men were diagnosed with prostate cancer, representing a cancer rate of less than 1% per year of treatment.

Prostate cancer rates were investigated among men with and without the prostatic precancerous lesion known as high-grade prostatic intraepithelial neoplasia (PIN). In this 12-month study 75 men with hypogonadism received TTh, including 55 men with benign pretreatment prostate biopsy, and 20 men with biopsy revealing PIN. A similar 12-month increase in PSA of 0.3 ng/mL was seen in both groups, corresponding to a 15% rise. A single cancer was detected, in the PIN group, representing an overall cancer rate of 1.3%. The 5% cancer rate among men with PIN compares with a 25% risk over 3 years in this population, suggesting no significantly increased cancer risk.

One study that examined the effect of TTh on PSA found that the overall change was mild, and the individual response varied considerably. Among 58 men who underwent TTh for 1 year, the majority (32 men) demonstrated a mild PSA increase of 0.5 ng/mL or less. There were also 14 men with a PSA increase greater than 0.5 ng/mL, but 12 men with a decline in PSA. No apparent differences in age, baseline T concentrations, or baseline PSA were noted between men with a PSA increase of greater than 0.5 ng/mL and men whose PSA declined. In 81 T-deficient men who received TTh, no increase was noted in PSA at 1-year intervals for up to 5 years.

To put these studies and their results in perspective, it is important to be aware that the observed PSA changes in multiple studies of approximately 15% to 20% is not much greater than the 13% increase noted over 1 year in 50- to 60-year-old men participating in the placebo arm of an unrelated study. In addition, the annual cancer rate of approximately 1% that shows up repeatedly in TTh trials compares favorably to cancer detection rates in men undergoing prostate cancer screening.

A meta-analysis of 19 placebo-controlled TTh trials found that men who received T had no greater risk of PCa than men who received placebo. There was also no greater risk of developing a PSA level of more than 4.0 ng/mL. Remarkably, administration of supraphysiologic doses of T for 9 months in healthy men resulted in no increase in prostate volume or PSA.

It is also worthwhile noting that two studies involving more than 400 men in total have shown that T-deficient men with PSA of 4.0 ng/mL or less have a biopsy-detectable cancer rate of 14% to 15%. This result means that 1 in 7 men with normal PSA and low T has PCa. If raising T truly caused PCa to grow more rapidly, one would expect TTh trials to be associated with high rates of cancer. The fact that this does not occur argues strongly that PCa risk is unrelated to serum T concentration.