What's known on the subject? and What does the study add?
Previous reports, with small numbers of patients, have described the problem of incomplete testosterone suppression (>1.1 or 1.7 nmol/L) with LHRH agonists. Various predisposing factors have been suggested: different drug agents and patient factors such as age, pretreatment testosterone levels and weight. Such incomplete testosterone suppression has been shown in one small report to be associated with increased PSA failure rates and in another report in those with metastases, with worse survival.
This study used testosterone assays that are more accurate at low levels than those used in most previous reports in a large dataset of 2196 men, and confirmed incomplete testosterone suppression (breakthrough) rates >1.7 nmol/L of 3.4% and >1.1 nmol/L of 6.6%. We showed that younger age was strongly associated with the risk of breakthrough, with a minor effect of increasing body mass index. Repeated breakthroughs were more common (16%) in those who had already had one breakthrough. Interim measures of cancer control (PSA kinetics during LHRH therapy) were inferior in those with a breakthrough, and those with breakthroughs between 1.1 and 1.7 nmol/L had worse long-term biochemical control rates.
• To describe breakthrough rates above castrate levels of testosterone, in a population-based series of men undergoing adjuvant luteinizing hormone-releasing hormone (LHRH) agonist therapy with curative radiation therapy.
• To explore the predisposing factors for such breakthroughs and their impact on subsequent outcomes.
PATIENTS AND METHODS
• All men treated for prostate cancer between 1998 and 2007 with curative radiation in the province of British Columbia, Canada were potentially eligible (n= 11 752). Of these, 2196 fulfilled the eligibility criteria.
• Serial testosterone measurements were obtained during continuous LHRH therapy.
• Breakthrough rates >1.1 nmol/L and >1.7 nmol/L were calculated for each LHRH injection and for each patient course.
• Predisposing factors were identified, and early surrogates of oncological outcome (neoadjuvant nadir and post-treatment nadir) were determined.
• The risk of a breakthrough >1.1 nmol/L was 6.6%, and >1.7 nmol/L was 3.4% per patient course and 5.4% and 2.2% per LHRH injection (inclusive ranges).
• Repeated breakthroughs occurred in 16% of patients.
• Younger men were more liable to breakthroughs (P < 0.001).
• Early PSA kinetic surrogates of cancer control were inferior in those with breakthroughs.
• Neither overall biochemical non-evidence of disease (bNED) nor survival were compromised, although subgroup analysis showed inferior 5-year bNED in those with breakthroughs of 1.1–1.7 nmol/L vs those without (58% vs 73%, respectively; P= 0.048).
• Breakthroughs with LHRH agonists occur occasionally per injection, but occur commonly per patient course of treatment, and adversely affect early surrogate measures of outcome.
• The monitoring of testosterone levels during therapy is therefore advised.
Recent discoveries concerning the androgen receptor (AR) show that intracellular and adrenal sources of testosterone may be a driver of so-called castrate-resistant prostate cancer . Inadequate suppression of testicular testosterone may provide further AR activation and it has been suggested that some of the efficacy of the newer agents targeting the AR may be attributable to ongoing testosterone activity . With this background, we undertook the present study of the completeness of testosterone suppression with LHRHa agonists. The ‘conventional’ castrate level for testosterone obtained with LHRHa is usually reported as <1.7 nmol/L (50 n/dL) , but >1.1 nmol/L PSA is no longer fully suppressed, according to one small report , and the testosterone level with surgical orchiectomy is much lower at <0.7 nmol/L . Testosterone levels with LHRHa may not achieve such levels, and higher testosterone levels have been associated with an increased death rate in those with castrate-resistant prostate cancer [4,5]. Previous reports, with small numbers of patients, have identified such LHRHa breakthroughs (also described as escapes) as being relatively common: e.g. cumulative breakthrough rates of 31.5% for 1.1–1.7 nmol/L and 24.7% for ≥1.7 nmol/L in a series of 73 patients ; however, most of these rates occurred at the 1-month time point, when initial LHRHa receptor stimulation (‘flare’) may still be occurring . Particular LHRHa formulations may differ in this phenomenon [8,9].
The present study was designed to explore the frequency of testosterone breakthroughs in a large population-based dataset so as to identify associated patient and pharmacological factors, and to investigate if breakthroughs affected subsequent outcome.
MATERIAL AND METHODS
The British Columbia Cancer Agency (BCCA) is the sole provider of radiation therapy to the residents of BC (population 4.5 million). The BC Cancer Registry is maintained by the BCCA and captures details of all patients diagnosed with malignant disease in the province.
The BCCA radiation warehouse table captures details on all radiotherapy delivered in the province. This was queried for all patients treated with radiation therapy with curative intent between 1998 and 2007 (n= 11 752). All approved anti-cancer medications are dispensed or reimbursed by the BCCA, and prescribing records are available electronically. The provincial pharmacy database contains records of LHRHa and anti-androgen prescriptions. Each record details the date of prescription, drug dispensed, dose and quantity. The start date of adjuvant androgen deprivation therapy (ADT), as well as stop date, and any use of secondary intervention for relapse were calculated. Gaps in prescribing were identified by comparing the sum of duration of each prescription with the total elapsed time of the ADT course. Four products were used: goserelin (Zoladex®, Astra-Zeneca, Mississauga, ON, Canada), leuprolide i.m. (Lupron®, Abbott Laboratories, Montréal, QC, Canada), buserelin (Suprefact®, Sanofi-Aventis, Laval, QC, Canada). Leuprolide s.c. (Eligard®, Sanofi-Aventis) also became available in 2005. The assumed duration of action of the products was taken from the drug monographs (i.e. 4 weeks for short-acting products, and 2–6 months for the longer-acting products). The BCCA Tumour Marker Laboratory is a reference laboratory for the province of British Columbia. During the era of this study several testosterone assays were in use: a manual radioimmunoassay (RIA) produced by Diagnostic Products Corp. (USA) from January 1989 until June 2004, and an automated chemiluminescent immunoassay (Elecsys) produced by Roche-Boehringer-Mannheim (Mannheim, Germany) from June 2004 to the present. In addition, an automated immunoassay produced by Abbott (Architect Chemiluminescent Microparticle Immunoassay [CMIA]) was used briefly from July 2001 until February 2002, when it was replaced by the RIA assay for results below 3 nmol/L. It had been determined that assays at such low levels may be unreliable, with overreading of the true result by factors of 2–5-fold , and therefore results with the CMIA assay and those from other laboratories (where it was usually not possible to reliably identify the testosterone assay used) at levels <3 nmol/L were discarded. Of a potential 5566 testosterone results (during ADT therapy for eligible patients) 89% were included according to these criteria.
The practice of frequent testosterone measures arose in BC, to allow us to more accurately carry out analysis of outcomes and be alerted to any patient whose testosterone profile indicated the use of an ADT agent. In practice, our ability to independently query the provincial pharmacy database made this unnecessary. Measurement of testosterone was performed at any time of day, and tended to coincide with a clinical appointment, not necessarily at the same time as LHRHa administration, as ‘home injection programmes’ are used by many patients, whereby the agent is administered outside the clinic.
Breakthrough levels of testosterone were a priori defined as (i) ≥1.7 nmol/L and (ii) ≥1.1 nmol/L. Because we did not want to investigate the flare phenomenon, any testosterone results obtained within 4 weeks of starting a course of LHRHa were excluded. PSA kinetics were examined as an early measure of subsequent outcome. Firstly, the neoadjuvant nadir PSA (that is, the minimum PSA level reached on ADT before initiation of radiation) has been shown to be a strong correlate of subsequent prostate cancer survival [11,12]. Secondly, the post-radiation nadir PSA is also of prognostic importance . Subsequent PSA outcomes were assessed using the Phoenix (nadir + 2) definition of relapse in a subgroup of patients treated with external beam radiation, excluding those participating in an ongoing randomized study, whose PSA results were not obtained , and those treated with brachytherapy.
Clinical data were obtained from chart review or from existing databases. Patients were eligible for the present study if they received an LHRHa with curative radiation therapy for prostate cancer, and had at least one testosterone measurement during that time. Systematic and random chart audits were performed, with 67% of all cases being audited. All patient charts with a breakthrough were reviewed to ensure data accuracy. Detailed exclusion reasons are given in Fig. 1. From the potential 11 752 patients, after the eligibility criteria were applied, 2196 patients were included. Ethical approval was given by the University of British Columbia to allow access to these data sources, and to carry out the present study.
The frequency of breakthrough was calculated both as the occurrence of breakthrough per patient (maximum where >1), and also per injection, where the LHRHa prescription had been accompanied by a valid testosterone measure. With the latter, the frequency of breakthroughs will be independent of the frequency of testosterone testing, but the rate per patient course will depend on the testosterone testing frequency. Patient factors studied for breakthrough risk were tumour risk group , age, body mass index (BMI) and pre-treatment testosterone, grouped according to Damber et al. . Treatment-related factors examined were the LHRHa type and year of treatment. These factors were analysed in relation to breakthroughs >1.1 nmol/L, using a generalized estimating equations logistic regression model to handle intra-subject variation (owing to multiple injections per patient). Analysis was conducted using SAS v9.2 (Cary, NC, USA; http://www.sas.com)
Patient and treatment characteristics are shown in Table 1. The median duration of ADT was 11.3 months and the median number of LHRHa injections per patient was 4. The most common product used was goserelin in 59% of cases, and the least common, leuprolide s.c., 5%.
Table 1. Patient and treatment characteristics
Pre-treatment testosterone intervals are divided according to Damber et al. .
Designated duration of LHRH injections, % of patients
Mean,no. of testosterone results per patient during ADT.
Median (range) no. of testosterone results per patient during ADT
Mean ratio of no. of testosterone measures to LHRH scripts:
The risk of a breakthrough >1.1 nmol/L was 5.4% and >1.7 nmol/L was 2.2% per LHRHa patient course. As patients typically received four injections each (36% of injections being accompanied by a testosterone measurement), the risk per patient course was higher: 6.6% and 3.4%. (ranges are inclusive). Table 2 contains more details of the testosterone measurement and breakthrough rates. When examined for the 3-month preparations, once established on the drug, breakthroughs appeared at a constant rate in relation to the time elapsed since injection, being neither more common just before or after each injection.
Table 2. Testosterone breakthrough details
Testosterone breakthrough rate per LHRH injection
1.1–1.7 nmol/L (%)
>1.7 nmol/L (%)
Highest recorded and breakthrough rate per patient
1.1–1.7 nmol/L (%)
>1.7 nmol/L (%)
Testosterone breakthrough rate by LHRH product (>1.1 nmol/L)
Leuprolide s.c. (%)
Leuprolide i.m. (%)
Repeated breakthroughs: frequency after first occurrence
1st breakthrough 1.1–1.6 nmol/L
1st breakthrough ≥1.7 nmol/L
2 breakthroughs (%)
3 breakthroughs (%)
≥4 breakthroughs (%)
Testosterone breakthrough frequency (≥1.1 nmol/L) by LHRH duration
4 weeks (%)
2 months (%)
3 months (%)
4 months (%)
6 months (%)
Repeated breakthroughs occurred in 16% of those with a first breakthrough >1.1 nmol/L, compared with a per-injection breakthrough rate of 5.4% (P < 0.001, chi-squared test). For those with a lower initial breakthrough (1.1–1.7 nmol/L) further breakthroughs occurred in 19%, compared with 11.5% in those with higher breakthroughs (>1.7 nmol/L). Of those where a breakthrough was noted by the oncologist (n= 20), seven of those with a breakthrough >1.7 nmol/L were switched to a different LHRHa, compared with none where the breakthrough was 1.1–1.7 nmol/L. Of the seven patients switched to a different LHRHa, castrate testosterone was obtained in three patients after switching, in two further patients after a further injection of the same drug before the switch had occurred, and in two patients still failing to achieve castrate levels with a different product.
Univariate analysis identified younger age as the only definite factor associated with breakthrough >1.1 nmol/L (P < 0.001, confirmed on multivariate analysis [Fig. 2]). Higher BMI, and lower risk group was of borderline significance and not confirmed on multivariate analysis (Table 3). Pre-intervention testosterone level was not associated with breakthrough risk.
With regard to the different LHRHa preparations, the risk of breakthrough >1.1 nmol/L per LHRHa injection appeared to be greater for the leuprolide s.c. (Eligard®) product, with an overall rate of 11% vs 5% for others, (P < 0.001 chi-squared test, Table 2). However, because this product was only prescribed in the last 3 years of our study when we had switched to the Elecsys testosterone assay, and because other unknown factors may have changed, multivariate analysis incorporating year and drug type, as well as those factors already shown to be significant, was undertaken. The overall hazard of breakthrough per patient with leuprolide s.c. was 1.7 (CI 0.79–3.74, P= 0.2) compared with leuprolide i.m. (Lupron®) and this was not statistically significant. There were no differences in the ADT duration, age or BMI in patients receiving leuprolide s.c. compared with other products. It should be noted that relatively few patients (5%) received leuprolide s.c., thus limiting the power of this analysis.
Patient outcomes, as assessed by early PSA kinetics, showed an adverse effect of testosterone breakthrough. The neoadjuvant nadir PSA was significantly higher in those with a breakthrough in the neoadjuvant period (median 0.68 ng/mL v 0.41 ng/mL; P < 0.001). The post-treatment nadir was also higher in those with breakthroughs: 0.02 ng/mL for those without a breakthrough, compared with 0.04 ng/mL for testosterone >1.1 nmol/L or >1.7 nmol/L (P= 0.008 and P= 0.003, respectively). Full details are given in Table 4. Subsequent biochemical non-evidence of disease (bNED) rates (i.e. biochemical control rates) after 5 years (with a median follow-up time of 45 months), was 73.1% in those without a testosterone breakthrough vs 62% with (P= 0.09). The subgroup of men with a breakthrough of 1.1–1.7 nmol/L, had a 5-year bNED rate of 57.6% (P= 0.048), and those with a breakthrough of >1.7 nmol/L, had a 5-year bNED rate of 68% (P= 0.6). It is noteworthy that, while 31% of those with breakthroughs >1.7 nmol/L had some action taken as a result of the breakthrough (most usually the addition of an anti-androgen), this only happened in 13% of the 1.1–1.7 nmol/L group patients, (P= 0.038, chi-squared test), probably because these breakthroughs were less pronounced and, therefore, either unrecognized or presumed to be of lesser importance.
Table 4. Effect of testosterone breakthrough on PSA kinetics
Neoadjuvant nadir PSA
Mean PSA (ng/ml)
Median PSA (ng/ml)
Significance (Kruskall-Wallis) v. no breakthrough
Breakthrough 1.1–1.7 nmol/L
P = 0.11
Breakthrough >1.7 nmol/L
P < 0.001
Post-treatment nadir PSA
Mean PSA (ng/ml)
Median PSA (ng/ml)
Significance (Mann Whitney U Test)
Breakthrough 1.1–1.7 nmol/L
P = 0.008
Breakthrough >1.7 nmol/L
P = 0.003
PSA control rate at 5 years (nadir + 2 definition)
Significance (log rank)
Breakthrough >1.1 nmol/L
P = 0.048
Breakthrough >1.7 nmol/L
P = 0.6
Previous series and case reports, most with fewer than 100 patients [3,5,17–21], have shown that failure to achieve castrate levels of testosterone <1.7 nmol/L occurs in up to 12% of patients [22,23]. The largest report (to date as abstract only) on 451 patients showed an 11% risk . The present series, from a population-based study of over 2000 men, confirms that breakthroughs occur with rates per patient of 3.4% ≥1.7 nmol/L and 6.6% ≥1.1 nmol/L.
Rates in the present study are lower than those in other reports because we restricted our analysis to those testosterone assays that have been shown to be more accurate at low levels; whereas other series have not done this and may therefore have overestimated the true rate. The rates taken from the drug monographs are typically 1–2% [25,26].
The conventional castrate level (1.7 nmol/L or 50 ng/dL) that is quoted in the literature was chosen based on testosterone assays that were insensitive below this level, but actual testosterone achieved with surgical castration is <0.7 nmol/L, according to the few studies examining this [4,27]. Although more modern testosterone assays have allowed the measurement of such low testosterone levels, it is not appreciated how poor many of the assays in current use may be. Ratios of measured to true results in women with testosterone values of ∼1.6 nmol/L showed overmeasurement by a factor of ≥3 in six of 10 assays studied . Many laboratories do not quote the assay used in their reports, nor do they provide guidance about interpretation of accuracy. The present study only included testosterone results at levels <3 nmol/L with the RIA and Elecsys assay, which had performed well in comparison with the ‘gold standard’ isotope dilution-liquid chromatography (IDLC)-tandem mass spectrometry . Few reports on testosterone breakthrough make mention of assay variability. A recent report of testosterone levels obtained with IDLC-tandem mass spectrometry showed that LHRHa agonists performed as well as orchiectomy, with only one of 32 subjects failing to achieve a testosterone of 0.7 nmol/L; however, assays were only performed once per patient, and the LHRHa preparation was not specified .
We found repeated breakthroughs occurred in 16% of those experiencing a first breakthrough, compared with an initial breakthrough rate of 5.4%. This suggests that breakthroughs are likely to be related to patient factors such as age and BMI, or are LHRHa-specific, rather than being random events. Similar results have been reported by Morote et al. , where the risk of subsequent breakthroughs was related to the magnitude of previous testosterone level on ADT. Those with 1, 2, or 3 consecutive readings <0.7 nmol/L had chances of 11%, 5.7% and 0%, respectively, of having a subsequent breakthrough >1.7 nmol/L. Most did not change to a different LHRHa as the breakthrough was not recognized by the treating oncologist, but, of those that did switch, 29% still failed to achieve low testosterone levels.
The large patient numbers in the present study permitted investigation into such potential predisposing factors. We identified younger age as a factor predisposing to breakthrough, which has previously been postulated , but the present study is the first to provide confirmation. The probable mechanism is a more robust hypothalamic-gonadal pathway in the young. The effect of increasing BMI was of only borderline statistical significance, but may be related to a dose/kg effect.
The patients in the present study received a variety of LHRHa products, and this allowed a direct, albeit non-randomized, comparison of breakthrough rate by product. Crude increased rates with leuprolide s.c. (Eligard®) were not confirmed in a multivariate model with no significant differences by product. Nonsignificantly increased rates for leuprolide s.c. were found in a large Spanish study (14.5% vs 10% for others, P= 0.22) . The randomized data that is referenced in the Eligard® product monograph quote breakthrough rates ≥1.74 nmol/L in 2% of patients with the 4-month preparation  and 1% with the 3-month and 6-month preparations [26,29], rates which are lower than the 4.4% rate per injection in the present study. Leuprolide s.c. has also been reported to fail to suppress testosterone according to Koupparis  and Koupparis and Tyrrell  perhaps owing to a granulomatous reaction to the microcapsules of the formulation, and others have suggested an autoimmune-mediated resistance . In Sweden, the Medical Protections Agency alerted physicians about 12 cases of failure of testosterone suppression with leuprolide  and consequently recommended close monitoring of testosterone while patients are on LHRHa. At least one author believes that ‘not all LHRHa agonists are equal’. A review of LHRHa efficacy compares breakthrough rates reported according to product  and shows that all products are prone to this phenomenon; apparent differences between products are difficult to interpret because of design differences in the studies reviewed.
While it may appear that the efficacy of intermittent androgen deprivation therapy (IADT) is inconsistent with the present results, we do not believe that it is. Firstly, the recently reported results of a large randomized comparison of IADT vs continuous therapy showed an increased prostate cancer death rate, suggesting inferiority for this endpoint (although quality of life was improved, and overall survival was similar) . Secondly, it has been shown that effective IADT requires a good PSA response, which in turn is dependent on an adequate castrate testosterone level , as reviewed by Shore and Crawford , who recommend ‘keeping testosterone levels as low as possible while patients are on treatment results in the lowest possible serum PSA levels and likely improves outcomes’. Likewise, anti-androgen monotherapy, a non-castrating treatment for advanced and metastatic prostate cancer, is established in European practice; however, survival is inferior compared with orchiectomy , but it is used owing to perceived quality-of-life advantages .
The limitations of the present study are its retrospective and non-randomized design, and the lack of standardized testosterone measurements during the ADT course. Confirmatory testosterone measures were generally only obtained when a higher level of breakthrough had been recognized by the oncologist, and more complete measures of endocrine function, such as gonadotrophins, were not measured. Another limitation is the inaccuracy of testosterone assays, which required 10.6% of all results to be discarded, but the strength of a registry-based and consequently large data source mitigates this to some degree. In addition, our pharmacy database records the date of dispensing of medication, with no guarantee that the drug was given on that date; however, all patient charts with breakthrough were audited, and where any unresolved concern about administration dates remained, the patients were removed from the study.
The breakthroughs described in the present report are associated with adverse effects on initial PSA kinetics, which have been shown to correlate with prostate cancer death on longer follow-up [13,37]. We did not detect a definite impact on subsequent PSA relapse (P= 0.09), although sub-group analysis did show that men with breakthroughs of 1.1–1.7 nmol/L (but not higher) had inferior biochemical outcomes. This may be because of intervention when the breakthrough occurred or, alternatively, the saturation theory of AR stimulation  is also compatible with these findings. The theory is that prostate cancer is exquisitely sensitive to testosterone levels in the near-castrate range, but becomes insensitive to higher testosterone levels. The present study was not designed to examine long-term PSA control rates and the relatively short PSA follow-up of 45 months may be insufficient to detect a difference if one truly does exist. There was no effect on prostate cancer death; however, again follow-up is short for this endpoint. In those at higher risk of relapse or death from prostate cancer, full castration may be of greater importance, with one retrospective report of 129 men showing the risk of death from castrate-resistant prostate cancer being related to the testosterone level after 6 months of LHRHa therapy .
If achieving true castrate testosterone levels is important, then testosterone should be monitored as recommended by several others [6,18,30] and, if necessary, a different LHRHa agonist or antagonist should be used.
In conclusion, we have confirmed that breakthroughs of testosterone suppression above conventional castrate levels of 1.7 nmol/L occur after 2.2% of LHRHa injections. Repeated breakthroughs are more than twice as common as initial breakthroughs, and breakthroughs >1.7 nmol/L per patient course were observed in up to 7.4% of patients, depending on the LHRHa used. There is an impact on early PSA kinetic outcomes, and possibly on longer-term biochemical control. It is prudent therefore to recommend that testosterone be monitored closely using an assay that has the required sensitivity.
Research Support: This project was the recipient of an ACURA (Abbott-Canadian Association of Radiation Oncology Uro-Radiation) award. We thank Jerusha Miller, Medical Student and the oncologists who cared for the patients included in this study. Dr Scott Tyldesley has a Career Investigator Award from the Michael Smith Foundation for Health Research.
CONFLICT OF INTEREST
Tom Pickles has received honoraria/speaker fees from Abbott Labs and Ferring Pharma. Scott Tyldesley has received honoraria/speaker fees from Astra-Zeneca and Sanofi.