Sunday, September 4, 2016

Testosterone to TREAT prostate cancer - are they crazy? No - it just may work. (mCRPC)

(frequently updated)


When prostate cancer metastasizes and becomes castration-resistant (mCRPC) after a period of androgen deprivation therapy (ADT), a number of biochemical changes to the androgen receptor (AR) take place within the cancer cell. Among those changes:
  • The androgen receptor (AR) multiplies on the cancer cell surface so that even the smallest amount of testosterone or other circulating androgens can activate it.
  • Even without an androgen ligand, the androgen receptor moves from the surface of the cell to the inside where it is protected. These internalized androgen receptors play a role in encouraging cell replication and in self-destruction (apoptosis) of cells in which the DNA  has become irreparably damaged.
  • The cell manufactures its own androgens internally. These activate the internalized androgen receptors.
  • The androgen receptor mutates into truncated versions that can be activated by a host of other molecules (other than testosterone), or don't require other molecules at all to activate it. Recently, researchers at Johns Hopkins identified a version called the "AR-V7 splice variant" that allows the cancer cell to multiply even when all androgens are completely eliminated. It is induced by long term androgen deprivation (see this link).
There may also be tissue-based effects. This means that, in a tumor, there are a variety of cell types. Castration resistance is not an all-or-none situation. Some cells within any given tumor will always remain hormone sensitive. Some cells, like cancer stem cells, have never been hormone sensitive. All of these cell types interact. Hormone-sensitive cells signal castration-resistant cells to become more like them. At the same time, castration-resistant cells signal hormone-sensitive cells to become more like them. Over time, the hormone-sensitive cells (being the ones that are killed by ADT)  lose the battle as the equilibrium shifts towards castration resistance.

How testosterone can help

Supraphysiologic doses, meaning serum levels greater than 1000 ng/dL, may help in several ways:
  • Testosterone prevents the androgen receptor from multiplying on the cell surface, and decreases the  number of androgen receptors already there, thereby increasing the cancer cell's sensitivity to subsequent androgen ablation (see this link). It also prevents the cell from becoming super-sensitized to androgens.
  • Inside the cell, high levels of testosterone prevent the internalized androgen receptors from encouraging cell replication - too much testosterone, as well as too little testosterone, discourages cancer cell replication (see this link).
  • High levels of testosterone and other androgens may induce damage the cancer cell's DNA (see this link). One of the interesting hypotheses is that combining high testosterone with a chemotherapy that is known to cause DNA damage will be particularly effective (see below).
  • Testosterone may be able to reverse the AR-V7 splice variant that is known to cause resistance to even second-line hormonal therapies like Zytiga and Xtandi (see this link).
  • On the tumor tissue level, testosterone supports the growth of hormone-sensitive cells while destroying castration-resistant cells, as described. This shifts the equilibrium back towards androgen deprivation sensitivity.
  • Selective Androgen Receptor Modulators (SARMs) may be able to provide similar benefits without some of the drawbacks (see this link).

Why can't we just give continuous high doses of testosterone?

It's been known for a while that testosterone plays a role in keeping healthy prostate cells healthy. We have observed that hypogonadal men (men who have low natural levels of testosterone) are more likely to have prostate cancer, and more virulent types. For a thorough recent review of this subject, see this link. In fact, one pilot study showed that men with mCRPC who had higher plasma testosterone levels (but still at castration levels) survived twice as long. They responded better to chemotherapy as well (see this link).

In 2009, Robert Liebowitz et al. reported on 96 patients who received very high dose testosterone replacement therapy (TRT) after some kind of prostate cancer treatment. For 59 of those men, the only treatment had been androgen deprivation. 12% were detectably metastatic. 60% had PSA progression while on TRT, but few had metastatic progression in that time period. For most of those, discontinuing TRT reversed the PSA progression.

There was a small (12 man) safety trial at Memorial Sloan Kettering. None of the men achieved supraphysiological serum testosterone levels from the transdermal gel they used. One man had to discontinue when he experienced spinal pain, but there weren't any other major side effects. Of the 12 men, 7 had decreased PSA (one had a 50% decrease). The other 5 had increased PSA, 2 by more than 50%. There was no regression of metastases in any of the men.

Other small trials of TRT alone had mixed results (see this link and this one).

So TRT alone doesn't seem to work. Both effects are necessary: (1) the TRT re-establishes hormone sensitivity, and then (2) the ADT kills off the hormone-sensitive cancer cells. For disease stabilization or regression, it seems to be necessary to alternate TRT with ADT. This is called bipolar androgen therapy (BAT).

I'm on intermittent hormone therapy - doesn't that accomplish the same thing?

Unfortunately, no. It had been hoped that intermittent ADT would accomplish two benefits:
  1. Delay the development of castration resistance, and thereby prolong survival, and
  2. Give men a break during which quality of life would improve temporarily
In fact, it accomplishes neither for most men. Intermittent ADT is sometimes inferior to continuous androgen ablation, perhaps especially for those with low metastatic burden. Castration resistance occurs at the same time with either intermittent or continuous ADT, and intermittent has not been shown to prolong survival.

After a long while on ADT, it takes a long time for a man's testicles to recover the ability to generate amounts of testosterone that are adequate to recover libido and help a man feel better. As far as the cancer is concerned, the cancer couldn't adapt if it were suddenly shocked by a large surge of testosterone. But during the "off-cycle,"as the amount of testosterone slowly increases, his cancer is able to adapt.  The cancer consequently thrives on the incremental increases rather than being killed by it. By the time the testosterone reaches a level that makes a measurable difference to his quality of life, the cancer has proliferated. His PSA has then gotten so high that the ADT "holiday" must be ended.

It sounds good in theory, but does it work in clinical practice?

Schweizer et al. conducted a pilot test at Johns Hopkins. They treated 16 asymptomatic men who were diagnosed a metastatic and castration-resistant. They were all still on ADT. They gave the men high doses of injected testosterone and treatment with the chemo drug etoposide (see above), which is normally ineffective in treating prostate cancer. After at least 3 cycles:
  • Half of them enjoyed a decline in PSA, most of those by more than a 50% decline
  • Half had a radiographic response, including 1 patient who had no discernable metastases
  • Half the patients did not respond at all, and PSA continued to rise
  • Responders had a higher pre-BAT PSA than non-responders
  • 10 of 10 patients (100%) responded to second-line ADT (Xtandi, Zytiga or Casodex) after BAT (some were resistant to those therapies before BAT)
  • 3 patients had severe toxicity attributable to etoposide.
It appears that BAT may at least delay progression in at least some men with mCRPC. It seems to resensitize their cancers to hormonal agents, even when it didn't succeed in decreasing PSA or evident (radiographic) progression. It does not increase survival (see the TRANSFORMER RCT below). However, the men periodically enjoyed enhanced quality of life from periodic high levels of testosterone. It's unclear whether etoposide chemotherapy added much to the response.

How can we tell who will respond and who will get worse?

Markowski et al. reported on 6 men treated with BAT. A PSMA PET scan (DCFPyL) 3 months into BAT treatment revealed that 3 of them had already progressed to having new metastases.

Before it can gain widespread use, is imperative that predictive biomarkers be found. So far, genomic analysis has failed to discover any.

Can it overcome resistance to Zytiga or Xtandi?

(Update 6/2017) Denmeade's group reported on 30 minimally symptomatic mCRPC men who had progressed on Xtandi.
  • 30% saw PSA reduced by at least 50%
  • 43% saw PSA increase from baseline; in 17%, PSA more than doubled.
  • 36% had some regression of disease
  • Median 8.6 months of radiographic/clinical progression-free survival
  • 54% responded to re-challenge with Xtandi with a subsequent drop in PSA by at least 50% and progression-free survival of 4.8 months
  • A third of those with the resistant AR-V7 mutation responded to BAT, and had lowered AR-V7 levels
  • 2 converted from AR-V7 positive to AR-V7 negative
  • There were adverse side effects while on BAT. Transient pain flare was the most common, affecting 40%. A few men suffered very serious side effects: pulmonary embolism, heart attack, urinary obstruction, gallstones and fatal sepsis.
(Update July 2020) In the RESTORE randomized clinical trial (RCT), the researchers are investigating whether BAT can restore sensitivity to Zytiga and Xtandi to mCRPC men who have already progressed while using those. They are excluding those who have already had chemotherapy, and no chemo is used in this trial.  They give monthly testosterone injections until there is radiographic progression. The RESTORE trial found that BAT was able to restore responsiveness to Xtandi but much less to Zytiga.
  • Following testosterone therapy, PSA declined by 50% in 30% of the group who'd previously taken Xtandi, and in 17% of the group who'd previously taken Zytiga. On these small groups, the difference wasn't statistically significant.
  • After BAT, PSA declined by ≥50% in 68% on rechallenge with Xtandi and lasted 13 months
  • After BAT, PSA declined by ≥50% in 16% on rechallenge with Zytiga and lasted 8 months
  • There was no benefit to rechallenge with either in men with the AR-V7 mutation
(Update December 2020) The "C-arm" of the RESTORE trial comprised 29 castration-resistant (mostly metastatic) patients who received only ADT but no second-line hormonal therapies. 
  • Only 4 men (14%) had a PSA decline ≥50% due to the testosterone therapy
  • Only 4 men had a reduction in their metastases. All of those had lymph node metastases only.
  • Testosterone therapy lasted for 9 months (median) before radiographic progression was detected
  • Maximum PSA response was achieved in 56 days. Only 7 patients had any PSA reduction.
  • PSA more than doubled in 52%, and increased markedly in 14% more.
  • 31% had some radiographic reduction of metastases.
  • Median radiographic progression-free survival was 8.5 months
  • Musculoskeletal pain was experienced by 40%
  • Other prevalent side effects were: hypertension (21%), breast tenderness (21%), leg swelling (17%), fatigue (14%), and difficulty breathing (10%). One patient died of a stroke.
  • 18 patients later received Zytiga or Xtandi, with excellent results.
  • There weren't any discernable genomic determinants of response.
Based on the RESTORE trial, this trial suggests that BAT should be reserved for patients who:
  1. have already progressed on Xtandi, or 
  2. a short duration of BAT may make the first use of Xtandi last longer. 
Both of those hypotheses should be tested in larger trials. It also brings into question whether PSA reduction should be used as an endpoint. PSMA PET/CT may be a better indicator of early progression on BAT (see this link). There is still no clue as to why only 31% respond.

In the TRANSFORMER RCT,  195 chemo-naive mCRPC men who failed therapy with Zytiga were randomized to receive either BAT or Xtandi. Everyone crossed over to the therapy they didn't previously get. With up to 2 years of follow-up:
  • Comparing BAT to Xtandi (before crossover), there were no significant differences in the time to clinical or radiographic progression (5.7 months in both groups) or reduction in PSA by ≥ 50% "PSA50" (28% and 25%)
  • After crossover, PSA50 was 78% for Zytiga->BAT->Xtandi vs 23% for Zytiga->Xtandi->BAT
  • After crossover, PSA-progression-free survival was 11 months for Zytiga->BAT->Xtandi vs 4 months for Zytiga->Xtandi->BAT
  • Those who went from Zytiga -> BAT -> Xtandi survived 37 months vs 29 months for those who went from Zytiga -> Xtandi (not significantly different)
  • BAT improved patient-evaluated quality of life
  • 3 patients received Keytruda after BAT and did very well. This observation led to the COMBAT trial using a different immune checkpoint inhibitor (Opdivo)
  • (update 5/7/24) Patients who never achieved serum T below 20 ng/dl did better with BAT than Xtandi.
Including BAT right after Zytiga failure and then using Xtandi increased the time to PSA progression, delayed radiographic progression, and reduced PSA. But waiting until after a failure on both drugs had no effect. This trial suggests that BAT may be useful after Zytiga but before Xtandi, particularly if T response to ADT has been less than optimal.

(Update Oct. 5, 2022) An exploratory analysis helps to explain why BAT has limited effectiveness (only 20-30% of  CRPC men derive any benefit from BAT), and what might be done to make it more effective. They focussed on the protein called c-MYC, which is known to be upregulated in advanced prostate cancer. They found:
  • High androgen receptor (AR) activity is required for BAT to work. Only about ⅓ of CRPC men have high AR activity.
  • But AR inhibition first is needed for T to raise its activity
  • High AR activity downregulates c-MYC. 
  • High doses of testosterone (T) increases AR activity.
  • Xtandi (but not Zytiga or other advanced antiandrogens) prevents acquired resistance to T because it upregulates the AR while it inhibits it.
AR activity (requiring tumor biopsy) may be a valuable biomarker.

They are running a clinical trial to test cycling between T and Xtandi after Zytiga failure.

(update January 2021) In the COMBAT trial, 45 heavily-pretreated men who were mCRPC and who have progressed on either Zytiga or Xtandi, received BAT followed by Opdivo (nivolumab - an immune checkpoint inhibitor).
  • ⅔ had at least some PSA reduction
  • 40% had a PSA reduction ≥ 50%
  • Median overall survival was 28 months
  • 24% had a measurable reduction in disease; 11% for 11 or more months
  • Median radiographic progression-free survival was 5.7 months
  • 1 patient had a complete PSA response
  • However, in 27% PSA got much worse after BAT
  • BAT seems to inhibit MYC - a genetic driver of prostate cancer
This trial suggests that BAT may prime prostate cancer cells so that they are more sensitive to checkpoint inhibitors.

(Updated 9/21/21) 30 heavily-pretreated men who were mCRPC and who have progressed on either Zytiga or Xtandi, received BAT and the PARP inhibitor olaparib. Half had DNA damage repair defects. This was hypothesized based on lab findings.
  • ¾ had at least some PSA reduction
  • 47% had a PSA reduction ≥ 50% (44% if 2 who dropped out for progression are added)
  • 23% had a 12-wk PSA increase ≥50% (28% if 2 who dropped out for progression are added)
  • Median progression-free survival was 14.8 months if they were mutation-free vs. 7.5 months if they had the defects.
  • 2 patients had a complete PSA response
  • 5 of 8 90+% responders were free of DNA damage repair defects
  • 3 of 6 90+% progressors had DNA damage repair defects
  • 5 patients had serious (grade ≥ 3) toxicity, including one death
This trial suggests that BAT + olaparib achieved good response regardless of DNA damage repair defect status. Excellent responders had mutations of the T53 gene or DNA repair genes (see this link). The cause of responder/non-responder differences remains elusive.

(update 6/3/2022) A small trial suggests that there may be genomic predictors. They found that some mCRPC patients got a better PSA response from BAT if they had germline (inherited) or somatic (tumor biopsy) genomic mutations of the tp53 gene and one of either RBI gene or PTEN gene loss. There are no clear indications - the results were mixed:
  • PSA decreased in 10 of 17 men with tp53 and PTEN loss
    • in 8, PSA decreased more than 50%
      • in 1, PSA became undetectable
  • PSA increased in 7 of 17 men with tp53 and PTEN loss
    • in 3 of those 7, PSA more than doubled
  • PSA decreased in 2 of 2 men with tp53 and RB1 loss
  • PSA increased in 3 of 3 men with RB1 and PTEN loss, so tp53 loss seems to be necessary for a BAT benefit
They also pointed out that in the recent trial of cabazitaxel+carboplatin at MD Anderson, the same genomic predictors of success were found. They propose a comparative trial of the 2 therapies in men who have mutations in tp53 and one of the two (either rb1 or pten).

(Update 11/8/2018) A new clinical trial at the University of Colorado, Denver will try transdermal testosterone instead of injections alternating with enzalutamide.

(update 6/2024) A new clinical trial at Roswell Park, Buffalo to restore sensitivity to ARSi.

(Update Jan 11, 2020) A new clinical trial at Johns Hopkins will combine BAT and Xofigo. Prior treatment with chemo and/or no more than one kind of second-line antiandrogen are allowed but not required. Recall that the RESTORE trial suggested that response was limited to lymph nodes only, so Xofigo may be a complementary treatment. Treatments that induce double strand breaks, like BAT and Xofigo, may be complementary, especially when combined with immunotherapy (see this link).

Other clinical trials include combining with Provenge, combining with Carboplatin, and using on patients who have genomic repair defects other than BRCA: ATM, CDK12, or CHEK2.

It will be important to determine, in future clinical trials, which men will respond to BAT and which men will not. Sadly, there may only be a survival benefit in a small subset of patients, although there is a quality of life benefit. Importantly, all clinical studies so far have only been Phase 2 trials in very small groups of patients. Larger trials will be necessary to prove safety and efficacy. There are already concerns about safety, so patients should not attempt BAT outside of a carefully monitored clinical trial.

There may be particular situations where BAT may be an effective therapy, but data are so far lacking: 
  • What, if any, is its benefit in men who are metastatic but still hormone-responsive (mHSPC)? (see this link)
  • Does it have any benefit in men who have already had docetaxel chemotherapy? 
  • Considering that metastases shrank in some men on BAT, should it be tried in symptomatic men as well? 
  • What is its effect on AR-V7-positive or negative men? 
  • Should it be used in combination with other therapies, such as PARP1 inhibitors, immunotherapies, and radiological therapies? 
  • What is the optimal sequencing of therapies? 
  • Is there any benefit to BAT  used along with radiation therapy in high-risk men (see this link)?


  1. The upcoming trial in Colorado excludes anyone who has been on Xtandi. So that will eliminate many applicants..

  2. I've read the book by Edward Friedman, PhD. His model is hard to follow but I think he may say that response to TRT will vary based on the relative abundance of membrane vs. intracellular androgen receptors and whether aromatase is being blocked. Supplementing with T also increases Estradiol which can be harmful. I think he may also say that the more cells have mutated the less predictable the outcome. I'd like to see a controlled experiment to test his theories.

    1. I have no idea why he would say that; estradiol has so far been beneficial in a controlled experiment:

    2. You can see links to some of his publications here:

    3. PATCH is a good trial.
      Friedman postulates that medium/high T plus estrogen is a harmful combo.
      PATCH is very high estrogen/castrate T.

      Friedman's hypothesis is supported by human observation and mouse studies. The studies that I have viewed were done after Friedman came to his conclusions.