Introduction “Cancer cells can succumb from too little or too much hormone” – Charles Huggins In the 1940s Charles Huggins made the groundbreaking discovery that surgical castration results in dramatic palliative benefits for most men with metastatic prostate cancer, and since then androgen deprivation therapy (ADT) has remained the backbone for treating advanced prostate cancers. While most people remember Dr. Huggins’ work evaluating the clinical effects of surgical castration in prostate cancer patients, he also realized early on that the flipside of the coin was also true – that administration of androgens can also result in an anti-tumor effect – and went as far as treating a handful of prostate cancer patients with testosterone injections . Through the years, researchers have revisited the concept of using testosterone to treat prostate cancer many times over, with generally mixed results [2–5]. However, preclinical models have consistently shown that high concentrations of androgens are able to suppress tumor growth in select models . These observations led Denmeade and Isaacs to propose a new paradigm for testosterone-based therapy in which cyclic intramuscular injections of high-dose testosterone are administered (ie, bipolar androgen therapy or BAT), with the goal of causing rapid fluctuations between low and high testosterone levels over a short period of time . Preclinical data Multiple preclinical studies have shown that both continuous and cyclic treatment with high-dose androgens inhibit the growth of prostate cancer cells – an effect most obvious in models with high androgen receptor (AR) expression [8–17]. Potential mechanisms of action explaining these paradoxical effects, include: induction of DNA damage, interfering with DNA replication, suppressing cMyc activity, inducing negative cell cycle regulators, and altering expression of genes involved in prostate cancer pathogenesis and drug resistance [8–17]. Of the aforementioned mechanisms of action, the induction of DNA damage is one of the better described and perhaps most poised to inform biomarker development and combinatorial treatment strategies. Studies have shown that rapidly transitioning from a castrate to high-androgen environment induces double stranded DNA (dsDNA) breaks, an effect likely mediated by topoisomerase IIB (TOP2B) [7,8,13]. As AR-mediated transcription proceeds, knots and tangles in DNA occur. In order to overcome these topologic constraints, TOP2B induces transient dsDNA breaks, which are then repaired by the enzyme. Supporting this model are preclinical data showing that the TOP2B poison etoposide is able to potentiate DNA breaks by inhibiting DNA re-ligation . In addition, in vivo experiments have demonstrated synergy when high-dose androgens are combined with DNA-damaging doses of radiation . Interestingly, DNA damage and apoptosis following exposure to testosterone appear to be transient, with resolution of DNA damage beginning within 24 hours. Clinical data As mentioned above, the concept of using testosterone to treat prostate cancer dates back decades, and contemporary studies testing physiologic doses of testosterone in men with castration-resistant prostate cancer (CRPC) have documented relatively poor clinical activity [18,19]. More recently, we developed a form of cyclic high-dose testosterone therapy we termed BAT. This approach was developed based on the rationale that supraphysiologic concentrations of androgens were necessary to realize the full clinical benefits of testosterone-based therapy. Clinically, BAT is administered as intramuscular injections of either testosterone enanthate or cypionate at a dose of 400 mg every 28 days. This dose and formulation was selected because it has been shown to consistently drive testosterone levels well into the supraphysiologic range (>1500 ng/dL), followed by a rapid decline to the near-castrate range at the end of a 28-day cycle [13,20]. We hypothesized that BAT would prevent AR levels from adapting to either high or low androgen conditions, thereby preventing the emergence of resistance. In addition, because testosterone-induced DNA breaks and apoptosis are transient, rapid cycling could result in repeated rounds of DNA damage, thereby enhancing antitumor effects . In our initial pilot study testing BAT, we treated 16 men with asymptomatic CRPC with BAT combined with etoposide for 3 months. Patients whose prostate-specific antigen (PSA) levels were falling at the end of this 3-month period were allowed to continue on BAT monotherapy until disease progression. In total, 14 patients completed the first 3 months of therapy and were response-evaluable. Overall, we observed high response rates, with PSA declines and tumor shrinkage on scans in 50% of patients . Treatment was well-tolerated and most serious adverse events were attributed to etoposide. Because most patients who responded to treatment continued to derive benefit during the BAT monotherapy phase, we ultimately abandoned combination BAT plus etoposide given that the added toxicity could not be justified. Although this was a small study, it did provide proof of concept that high-dose testosterone was safe and effective for at least some men, and ultimately paved the way for a series of larger studies aimed at better defining the clinical effects of BAT. The next study testing BAT enrolled men with newly diagnosed hormone-sensitive prostate cancer . This study involved a 6-month lead-in period of ADT, with the goal being to drive AR levels up so as to ‘prime’ patients to respond to BAT. Patients then went on to receive alternating cycles of BAT followed by ADT for a total of 1 year. In that trial, we found that BAT resulted in high rates of PSA suppression at the end of the study, with improvements in quality of life. Given that this was a single-arm study, however, it remains unclear if BAT would prove a viable treatment option in this clinical disease space, especially in light of data showing that more aggressive inhibition of AR signaling with abiraterone dramatically improved survival in this setting. As such, we have refocused our attention toward testing BAT in men with CRPC. More recently, we published results from the RESTORE trial, a phase II study testing BAT in asymptomatic CRPC patients progressing on enzalutamide (n=30). This study had two goals: 1) to evaluate response rates in a population of patients post-enzalutamide; and 2) to assess responses upon rechallenge with enzalutamide. The later objective was included on the basis of data from our pilot study showing that post-BAT response rates to AR-targeted therapies were higher than expected, suggesting that BAT could combat acquired mechanisms of resistance to drugs like enzalutamide. In addition, preclinical models have shown that testosterone is able to suppress two key mediators of resistance to ADT: full-length AR expression and constitutively active AR splice variant expression. The RESTORE study met its primary endpoint, with 30% of patients demonstrating PSA declines ≥50% from baseline to BAT, and 52% of patients with a PSA decline ≥50% following re-challenge with enzalutamide [22,23]. This study also found that BAT was associated with improvements in quality of life. It was not clear whether suppression of full-length AR or AR splice variants explained this ‘resensitization’ to enzalutamide, however. Conclusions Correlative studies aimed at understanding the biologic factors influencing BAT’s clinical effects remain a major gap in developing testosterone-based therapies. Anecdotal reports have documented an extreme response in a man with impaired homologous recombination, and it stands to reason that the presence of impaired DNA damage-repair pathways may result in enhanced clinical effects if induction of DNA damage is BAT’s prevailing mechanism of action. Indeed, we will be opening a prospective study based on this concept, testing BAT in combination with the PARP inhibitor olaparib [clinicaltrials.gov: NCT03516812]. Importantly, this study will not require the presence of DNA damage-repair deficiency in order to enroll, although all tumors will be tested to assess for mutations in homologous recombination genes. As it stands, available studies support the further development of BAT. It is important to note, however, that BAT remains an experimental therapy. Patients included in all of the aforementioned studies have been highly selected, including the requirement that they have asymptomatic disease, lest BAT induce a tumor flare and lead to unacceptable toxicity (eg, cord compression, urinary outlet obstruction). We eagerly await the results of the randomized phase II study testing BAT versus enzalutamide in patients previously progressing on abiraterone (clinicaltrials.gov: NCT02286921), as this will better define the clinical effects of BAT and its role in treating late-stage prostate cancer. Until then, enthusiasm should be tempered and providers should be discouraged from rushing to incorporate BAT into clinical practice.