Mass spectrometry redefines optimal testosterone thresholds in prostate cancer patients undergoing androgen deprivation therapy

Mass spectrometry redefines optimal testosterone thresholds in prostate cancer patients undergoing androgen deprivation therapy - Beck - The Prostate - Wiley Online Library

1 INTRODUCTION

Noncurative treatment of advanced prostate cancer (PCa) relies mainly on the use of androgen deprivation therapy (ADT). Despite ADT efficacy, PCa will eventually progress to castration-resistance prostate cancer (CRPC) but many will remain dependent on steroids for growth. This explains why the combination of ADT with androgen receptor axis-targeted therapies (ARAT; apalutamide, darolutamide, enzalutamide, abiraterone acetate) delayed time to progression and death in metastatic PCa or nonmetastatic CRPC.^1-7 For patients undergoing ADT for nonmetastatic hormone sensitive PCa (mHSPC), the benefits of adding an ARAT has not been shown. Therefore, close monitoring of testosterone levels is indicated to ensure maximal AR axis blockade. A testosterone castration threshold of 0.7 nM has been suggested based on the testosterone level measured by immunoassays (IA) in PCa patients who underwent bilateral orchiectomy (median: 0.5 nM; 95% confidence interval [CI]: 0.42–0.60).⁸ Since then, it was demonstrated that a castrated testosterone level below 0.7 nM is associated with longer time to castration resistance and increased overall survival, suggesting that patients harboring higher testosterone level under ADT are at higher risk of progression.^9-15 However, most of these studies were performed using testosterone measurements by IA.^9-15 Considering that testosterone measurement methods have recently evolved, a redefinition of the optimal testosterone castration level using contemporary measurement methods is required to identify PCa patients at high-risk of progression.¹⁶

In this study, we have determined and compared serum testosterone level measured by IA and mass spectrometry (MS) in a cohort of PCa patients undergoing continuous ADT. Moreover, we correlated the time to CRPC with testosterone level to redefine optimal testosterone target threshold based on MS.

2 MATERIAL AND METHODS

2.1 Study cohort

This study was approved by the CHU de Québec-Université Laval institutional review board (2020-4900) and performed in accordance with the Declaration of Helsinki. This retrospective study investigated serum testosterone measurements performed between 2015 and 2019 at the CHU de Québec-Université Laval. Of the 412 PCa patients initially identified, 138 were undergoing noncurative continuous ADT and were included in this study (Figure 1). For 108 serum samples, testosterone measurement was performed by the Roche Diagnostics electrochemiluminescent IA and using LC-MS/MS adapted from a published method.^{17, 18} Supporting Information: Table 1 lists the clinical and pathological characteristics of the study cohort. The primary outcome was time to CRPC progression, defined as: (1) three prostate-specific antigen (PSA) increases (≥25%) with a minimal PSA value of 1 ng/mL or, (2) PSA level was below 1 ng/mL and either radiological progression or PCa treatment intensification.

Details are in the caption following the image — Flowchart of the patients included in the cohort study. The initial cohort consisted of 412 prostate cancer patients for which testosterone measurement assessed by mass spectrometry (MS) was available. Depicted are the 274 patients excluded from the study, leaving 138 prostate cancer patients undergoing noncurative continuous androgen deprivation therapy (ADT) included in the final analyses. Paired testosterone measurement by immunoassay (IA) was available for 108 samples. CRPC, castration-resistant prostate cancer; n, number.

2.2 Serum testosterone measurement by LC-MS/MS and IA

Serum testosterone levels were measured as previously described.¹⁶ Briefly, for MS measurements, the standards, quality controls and samples were prepared as described and injected into the Acquity ultrahigh-pressure liquid chromatography (LC) and online solid phase extraction system (Waters) using partial loop mode. After chromatographic separation of testosterone from other components, the eluate was injected into the XEVO TQ MS tandem mass spectrometer (Waters) as described.^{16, 17} The lower limit of quantification (LLOQ) was established at 0.1 nM.

IA testosterone measurements were performed using an electrochemiluminescent IA on the automated modular platform from Roche Diagnostics with an LLOQ established at 0.416 nM.¹⁸

2.3 Statistical analysis

Testosterone measurements below the LLOQ for both IA and MS methods were replaced by half of the LLOQ for each method corresponding to 0.05 nM (MS) and 0.208 nM (IA). Nonparametric (Mann-Whitney test; two-tailed) test was used to compare the distribution of testosterone measurements of patients who progressed to CRPC at the end of our study to patients who did not progress. The Contal and O'Quigley method was used to estimate the optimal testosterone castration level for IA and MS methods for the dichotomized clinical outcome (CRPC: yes vs. no) and the time-to-event data (months). Optimal cut-off point was selected based on maximizing the Q-statistic value and statistical significance. Binary variable below or at and higher than the identified cut-off point for each testosterone measurement method were created, and the outcome variable analyzed. Survival curves were generated using the Kaplan-Meier analysis. Univariate and multivariate Cox regression analysis were used to evaluate the relationship between testosterone threshold level and progression to CRPC. Confounder variables included in the multivariate analysis were the International Society of Urological Pathology grade group at biopsy (1 vs. 2 vs. 3 vs. ≥4), metastatic status at the beginning of ADT (M0 vs. M1) and PSA at diagnosis (≤10 vs. 10–20 vs. ≥20 ng/mL). All analyses were performed using the GraphPad Prism 8 software and SAS (version 9.4).

3 RESULTS

3.1 Testosterone level in PCa patients undergoing ADT as an indicator of progression to CRPC

Out of the 138 serum samples derived from PCa patients undergoing noncurative and continuous ADT for which testosterone was measured by MS, 108 samples had paired IA testosterone measurement available (Figure 1). Mean testosterone value for both measurement methods was below the clinical threshold previously established (≤0.7 nM) for PCa patients undergoing ADT (IA: 0.370 ± 0.281 nM; MS: 0.275 ± 0.204 nM) (Supporting Information: Table 1). Indeed, in our cohort, 84.2% (n = 91/108) of samples had a testosterone level below 0.7 nM when assessed by IA compared to 97.1% (n = 134/138) by MS.

At the time of analysis, 78.7% of patients progressed to CRPC in the IA cohort (n = 85/108) and 76.8% in the MS cohort (n = 106/138) with a median follow-up of 58.8 and 54.7 months, respectively. No significant difference in testosterone levels assessed by IA was seen between patients who progressed and those who did not at the time of our analysis (0.387 ± 0.297 nM vs. 0.305 ± 0.204 nM; p = 0.277) (Figure 2). However, measurement by MS in the same cohort (n = 108; paired IA samples) showed that patients who were to progress to CRPC had, a higher testosterone level (0.308 ± 0.221 nM) compared to patients who did not progress (0.214 ± 0.179 nM; p = 0.011) (Figure 2). This difference remained significant when we analyzed the full MS cohort (n = 138) (0.294 ± 0.212 nM vs. 0.213 ± 0.159 nM; p = 0.018) (Supporting Information: Figure 1).

3.2 Determination of the optimal testosterone threshold to predict the risk of progression to CRPC

Considering that eventually almost all PCa patients undergoing continuous ADT will progress to CRPC, biomarkers for the prediction of this outcome must also consider the time-to-event variable. Therefore, we used the Contal and O'Quigley method which uses both the CRPC outcome (0 vs. 1) and the time to CRPC variables by calculating the Absolute Log-rank value for each testosterone level to determine the optimal testosterone threshold to predict progression to CRPC. Based on this analysis, a testosterone level equal or higher than 0.705 nM as assessed by IA (Absolute Log-rank = 4.993, Q-statistic = 0.765, p = 0.300) and 0.270 nM for measurement by MS (Absolute Log-rank = 13.224, Q-statistic = 1.960, p = 0.0009) were identified as the optimal cut-off points to predict CRPC (Figure 3).

Using biochemical, clinical, and radiological criteria to assess progression to CRPC, we evaluated the prediction performance at 24 months of both testosterone measurement methods, 24 months being the median time to CRPC in our cohort (24.7 months; 95% CI: 20.1–29.2) (Table 1). The patients included in this analysis (IA, n = 64; MS, n = 84) had at least 24 months of follow-up or progressed to CRPC during the first 24 months of their ADT. Prediction of progression to CRPC at 24 months using the IA 0.705 nM cut-off point led to an ideal specificity and positive predictive value (100% for both) compared to MS cut-off point (72.22% and 86.84%, respectively) (Table 1). These results can be explained by the low number of patients having a testosterone level higher than 0.7 nM in this analysis (IA: 9 out of 64 patients), and as indicated by the higher false-negative rate using IA (82.35%) compared to MS (50.00%). The MS cut-off point of 0.27 nM showed a higher sensitivity to predict CRPC compared to IA 0.705 nM (50.00% vs. 17.65%). These results indicate that the MS testosterone cut-off point is more accurate to identify patients who will progress to CRPC (MS: 54.76% vs. IA: 34.38%) (Table 1).

Table 1. Castration-resistant prostate cancer prediction performance at 24 months of lower testosterone level cut-off point assessed by IA and MS.

	Testo IA ≥ 0.705 nM (%)	Testo MS ≥ 0.27 nM (%)
Sensitivity	17.65	50.00
Specificity	100.00	72.22
PPV	100.00	86.84
NPV	23.64	28.26
Accuracy	34.38	54.76
False positive	0.00	27.78
False negative	82.35	50.00

Note: Performance was assessed using samples from patients for which we had at least 24 months of follow-up after androgen deprivation therapy induction or progressed to CRPC during this period. Other samples were censored for this analysis.
Abbreviations: IA, immunoassay; MS, mass spectrometry; NPV, negative predictive value; PPV, positive predictive value.

3.3 Prediction of progression to CRPC using a lower testosterone level cut-off point

Based on the prediction performance analysis results, we validated the association of the identified testosterone threshold with time to CRPC using survival analyses. Kaplan Meier analysis indicated a nonsignificant difference in median time to CRPC when using testosterone measurement by IA equal or higher than 0.705 nM (16 months; 95% CI: 11–34 months) compared to testosterone <0.705 nM (26 months; 95% CI: 15–33 months; p = 0.0973) (Figure 4A). In univariate (hazard ratio [HR]: 1.579; 95% CI: 0.908–2.745; p = 0.1058) and multivariate (HR: 1.633; 95% CI: 0.780–3.418; p = 0.194) analysis, testosterone measurement by IA ≥ 0.705 nM was not associated with an increased risk to develop CRPC (Figure 4B).

Using MS testosterone measurement, Kaplan Meier analysis shows that testosterone ≥0.27 nM is associated with a shorter time to CRPC (15 months; 95% CI: 12–27 months) compared to testosterone <0.27 nM (26 months; 95% CI: 28–37 months; p = 0.0055) (Figure 4C). Cox regression analysis demonstrated that testosterone level assessed by MS ≥ 0.27 nM is associated with an increased risk of progression to CRPC in univariate (HR: 1.717; 95% CI: 1.160–2.541; p = 0.0069) and multivariate analysis, after correction for metastatic status, ISUP grade and PSA at diagnosis (HR: 1.662; 95% CI: 1.043–2.648; p = 0.033) compared to MS testosterone <0.27 nM (Figure 4D).

4 DISCUSSION

This retrospective analysis evaluated the association of a lower castrated testosterone threshold derived from MS method with progression to CRPC. The newly identified optimal testosterone cut-off point for PCa patients under ADT using MS measurement was of 0.27 nM compared to 0.705 nM for measurement by IA. The MS cut-off point was more sensitive and accurate to predict progression to CRPC compared to IA. A MS testosterone level ≥0.27 nM was significantly associated with a shorter time to CRPC and an increased risk to develop CRPC in univariate and multivariate analysis. Overall, our results redefine testosterone threshold of less than 0.27 nM determined by MS to identify hormone-sensitive PCa patients under ADT at higher risk of progression to CRPC and who may benefit from ARAT.

Very low circulating testosterone levels are found in PCa patients under ADT, therefore a sensitive and specific measurement method is required. As depicted in our cohort, 70% (76/108 patients) had an IA testosterone level below the LLOQ (0.416 nM) and demonstrates that IA's low sensitivity prevents the accurate evaluation of testosterone levels in castrated PCa patients. Mean of IA testosterone measurements in our cohort is below the LLOQ given that data below the LLOQ were replaced by IA's half of the LLOQ value (0.208 nM). This also translates in a very high percentage of patients having a testosterone level below the previously identified castration threshold (0.7 nM) as assessed by IA (84.2%) and MS (97.1%) in our cohort. The difference in the measured castration efficacy between the two methods is explained in part by the lack of specificity of IA.¹⁶ We have previously shown that 50% of IA testosterone level measurement >0.7 nM were <0.7 nM when measured by MS.¹⁶ Other studies have also shown that between 60% and 91% of PCa patients under ADT have a testosterone level below 0.7 nM, mainly using IA measurements.^{8, 11, 19} Although these patients show improved clinical outcomes, the identification of this threshold was mainly based on testosterone measurement by IA after bilateral orchiectomy.⁸ Modern testosterone measurement methods have increased sensitivity. Therefore a more systematic analysis of the optimal testosterone castration threshold is required. For that purpose, we performed the Contal and O'Quigley analysis and evaluated the association of each testosterone level with progression to CRPC. Morote et al. performed a similar analysis resulting in the identification of an optimal testosterone castration threshold of 1.12 nM, using IA measurement (n = 73).¹² On the other hand, our analysis was performed using both IA and MS measurement methods, where the LLOQ of our methods are lower (IA: 0.416 nM, MS: 0.1 nM) than the one of the IA method (0.525 nM) used in this previous study, which could explain the discrepancy.¹² Interestingly, in our study, the testosterone level as measured by MS in PCa patients who underwent bilateral orchiectomy (n = 5) was of 0.29 nM, similar to the MS cut-off point identified (0.27 nM). This supports that the IA cut-off point previously identified in 2000 by Oefelein (0.7 nM) in patients who underwent an orchiectomy was due to the measurement method available at that moment.⁸

The optimal testosterone castration cut-off point needs to be highly sensitive and accurate to predict CRPC, to identify most patients at risk of castration resistance. The prediction performance of the identified IA and MS cut-off points depicted a better specificity and PPV for IA to predict CRPC compared to MS. This can be explained by the low number of patients having a testosterone level higher than 0.7 nM in this analysis (n = 9 by IA), as indicated by the higher false negative rate of IA compared to MS to predict CRPC at 24 months. On the other hand, MS testosterone threshold was more accurate than IA (Table 1 and Supporting information: Figure 2). Prediction of CRPC using MS testosterone cut-off point leads to a false positive rate of about 30% which could lead to overtreatment of these patients and undue associated side effects if an ARAT is added.

Testosterone level higher than the classical 0.7 nM threshold is known to be associated with shorter time to CRPC and worse overall survival.¹¹ Studies have also suggested that higher baseline testosterone levels could be associated with improved clinical outcomes in CRPC patients undergoing ARAT combined with ADT.^20-24 Knowing that androgens and particularly the potent AR activator testosterone still play a crucial role in the progression of CRPC, there is an opportunity to identify nonmetastatic PCa patients with higher testosterone levels on ADT to consider treatment intensification with ARAT. On contrary, our results might also suggest that some patients that are candidate to ARAT (mHSPC) or node positive candidate to radiation therapy might not benefit from treatment intensification if their testosterone level by MS under ADT is <0.3 nM. A post hoc analysis of the prospective phase-III studies in this patient population could answer this question.^{2, 3} Indeed, our results suggest that a lower castrated testosterone threshold could help to better stratify PCa patients undergoing ADT.

We would like to acknowledge that our study has limitations. This study was a retrospective analysis, so testosterone sampling time was not pre-defined in ADT cycles. Also, the testosterone levels below the LLOQ have been approximated to 1/2 LLOQ. Previous studies have also used zero or LLOQ values which could provide slightly different T levels cutoffs to predict CRPC. Moreover, the determined testosterone threshold of 0.27 nM would need to be externally validated in an independent cohort, ideally from a Phase-III trial. In addition, the decreased performance of our IA methods may not apply to new IA testosterone measurement methods that are more sensitive that the one used in our study.²⁵ Nevertheless, we used the gold standard testosterone measurement method, namely MS, to assess the optimal testosterone castration threshold. In addition, both IA and MS testosterone measurements were all performed at the same institution, reducing the inter-platform variability known to cause great variability in testosterone measurements.²⁶

The results of our study support the importance to use highly sensitive and specific testosterone measurement methods for the assessment of ADT efficacy. Testosterone levels ≥0.27 nM performed using less sensitive methods should be further confirmed by either MS or IA methods validated for low testosterone levels. Persistent testosterone levels above the identified threshold could indicate that these patients may benefit from a change of ADT agent or from the addition of ARAT to delay progression to CRPC.

5 CONCLUSION

Mass spectrometry redefines a lower testosterone castration threshold to identify PCa patients undergoing ADT that are at high risk of earlier castration resistance. These patients could benefit from treatment intensification with agents like ARAT.

AUTHOR CONTRIBUTIONS

Study concept and design: Dominique Guérette and Frédéric Pouliot. Acquisition of data: Jérémie Beck, Francis Lemire, Michel Déry, Benoît Thériault, Gabriel Dubois, and Dominique Guérette. Analysis and interpretation of data: Jérémie Beck, Mélanie Rouleau, Bertrand Neveu, Francis Lemire, Dominique Guérette, and Frédéric Pouliot. Drafting of the manuscript: Mélanie Rouleau, Jérémie Beck, Bertrand Neveu, Dominique Guérette, and Frédéric Pouliot. Critical revision of the manuscript for important intellectual content: Jérémie Beck, Mélanie Rouleau, Bertrand Neveu, Francis Lemire, Michel Déry, Benoît Thériault, Gabriel Dubois, Dominique Guérette, Frédéric Pouliot. Statistical analysis: Mélanie Rouleau and Jérémie Beck. Obtaining funding: Frédéric Pouliot. Administrative, technical or material support: None. Supervision: Dominique Guérette and Frédéric Pouliot.

ACKNOWLEDGMENTS

The authors thank the CHU de Québec-Université Laval biostatistical service platform, particularly Narcisse Ulrich Singbo. This work was supported by a grant from the Fonds de Recherche du Québec-Santé (FRQS) for clinician-scientists (FRQS-35020).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Search This Blog

Informed Prostate Cancer Patient Support