Treatment of Metastatic Castration-resistant Prostate Cancer Patients With Abiraterone Acetate and Prednisone and Corresponding Survival Prognostic Factors

Discussion

In this study, a good response to treatment with AA+P was observed within the studied series of patients suffering from mCRPC. Likewise, a large amount of patient clinical data were studied, differentiating patient characteristics and those pertaining to the disease itself linked to higher survival rates.

The pivotal trial COU-AA-302 confirmed the efficiency of using AA+P treatment for patients with mCRPC who had not been previously treated with chemotherapy. It significantly improved the overall survival rate, and radiological progression-free survival (rPFS), maintaining an acceptable profile of secondary effects. The mean survival observed in the cohort studied here was below that of the pivotal trial (3-5) (24 months and 34.7 months, respectively). In part, this could be due to the different baseline characteristics of the patients included in both studies. The fact that patients who already presented with metastases were included in the present study - which implies higher tumour incidence and therefore a worse prognosis - could explain the poorer results in terms of survival. Similarly, the fact that the patients included in the present study had already symptoms and suffered from a worse functional status (ECOG 2) - who had specifically been excluded from COU-AA-302 - could justify the aforementioned lower survival results. Likewise, these patients were treated with fewer subsequent lines of AA+P treatment after progression. The poorer baseline status and the higher tumour incidence presented by some of the patients studied, made the subsequent use of more treatment lines less probable (67% of patients within the COU-AA 302, and 30% in the present case). However, these two studies must be compared with caution, given the differences in terms of study design and sample size.

However, other recent studies of Asian patients with mCRPC who were treated with AA+P revealed a mean survival of 23.7 (10) and 18.1 months (11), which are similar to the results obtained in the present study. It is important to highlight the interest of the survival results presented by Boegemann et al. (2), which showed a mean survival of 23.3 months. This retrospective observational study included a selection of 481 patients suffering from mCRPC from four European countries. All these patients had not been previously treated with chemotherapy, had received AA+P treatment, and had similar baseline characteristics to those presented in the present study: an average of 75 years of age, and functional baseline status (ECOG 0-1 in 87%, versus 92.4% in the present series). However, the patients in the present study presented with PSA, AP and LDH baseline values which were lower at the start of treatment: (PSA: 20.30 ng/ml vs. 56.2 ng/ml, AP: 90 UI/l vs. 119 UI/l and LDH: 178 UI/l vs. 277 UI/l, respectively). Equally, in the present study the proportion of patients found with initial metastases was higher (49% vs. 28.9%). The initial objective of Boegemann’s group was to determine the time to treatment faillure, meaning the interruption of drug administration due to the progression of the disease, intolerance to the treatment, or death. A mean time of 10 months was observed, versus the 16 months obtained in this study. However, the survival mean was similar in both studies.

As was to be expected, known factors that predict overall survival in mCRPC were also observed in this study, such as the duration of time between disease spread and the PC diagnosis (12), pain intensity measured by the BPI scale (13) and LDH serum levels (14, 15). Additionally, the duration of time between the initial PC diagnosis and the start of AA+P treatment, together with the characteristics the disease at diagnosis, were other variables linked to survival in the present series of patients. On the contrary, age, score within the ECOG scale at the start of AA+P treatment, longer ADT response or treatment received once the patient was diagnosed with PC, the ISUP grade, number of bone lesions, or AP and PSA base line serum levels did not significantly impact statistics within the present series of patients, while they are known prognostic factors in other studies.

In keeping with other published studies, the results obtained here point to the possibility that certain clinical parameters can predict patient response to AA+P treatment. With this objective in mind, a variable was incorporated that divided the present patients into risk groups, depending on the stage presented at the time of PC diagnosis. Thus, patients with localised (low, intermediate and high risk), locally advanced and metastatic PC were included. Some of these patients underwent treatments with curative intent (surgery, RT); others, were treated with the combinartion of RT and ADT; whilst others were treated only with ADT, prompted by a clinical decision, because at the time of diagnosis, bone metastasis was already present, or because at the time it was the only available treatment for the disease. It is, therefore, logical to infer, that the evolution of the disease until the castration-resistant phase is reached would differ in each of these patient sub-groups. This can be attributed to tumour incidence, and to the treatment or lines of treatment received until disease progression.

The results obtained here suggest a lower benefit derived from long-term AA+P treatment in patients with high initial metastatic incidence, who, until the disease progressed, had only been treated with ADT. Nonetheless, the LATITUDE analysis (16) in patients with high-risk metastatic castration-sensitive PC, supports the addition of AA+P with ADT compared with ADT alone. This was in terms of survival benefits, by significantly increasing their overall survival rate and rPFS.

In this study, the patients with the longer survival, showed a longer time span between initial PC diagnosis and the development of castration resistance, followed by the consequent start of AA+P treatment. Hence, a duration of over 45 months was significantly linked to longer survival. Contrary to what was expected, ADT time was not linked to longersurvival of patients in the present work. Numerous studies have confirmed that a shorter response time to ADT is linked to lower PFS and overall survival rates (17). In keeping with this, Fan et al. (10) observed that longer duration to ADT response (≥18 months) was a determining factor for the survival of patients treated with AA+P. They suggested that patients administered ADT during a short period of time until castration resistance is developed (<18 months), could benefit from initial systemic chemotherapy treatment. Within the previously mentioned Boegemann et al. (2) study, the patients who had longer time between the start of their hormone deprivation treatment and the development of castration resistance, benefited from more time until treatment failure was reached. The present study included metastatic patients who were initially treated with ADT; the disease at this stage intrinsically implies high tumour burden, with a high probability of tumour behaving aggressively, which in turn impacts the value of the ADT time variable found in the present series of patients.

Hence, patients experiencing lower pain intensity at the start of AA+P treatment (BPI 0-2), show higher survival probability. This is in keeping with a post-hoc analysis of the COU-AA-302 study (18), which showed that the clinical benefits of AA+P treatment are higher for patients who had less pain at the start of treatment (BPI 0-1). Armstrong et al. (19) analysed patients from the Prevail study data base, with the objective of defining prognostic and predictive variables for overall survival rates. They validated a prognostic model which included, among others, 11 pain intensity variables at treatment onset using Enzalutamide. They found that the lower the baseline pain, the higher the survival rates.

The majority of the tumour biomarkers studied carried prognostic value at the time of diagnosis, since their concentration was related to the size of the tumour (20). Their true clinical value, however, resides in patient follow-up, to detect early relapse and also to evaluate the effectiveness of the prescribed treatment. Within the broad range of markers found in literature, some have been validated as useful serum markers from the prognostic point of view, and that is why PSA, LDH and AP are included in the present analytical determinations.

LDH is an important marker of swelling and tissue damage (21). In this study, the LDH serum concentration at the start of AA+P treatment was significantly higher in patients with shorter survival. Using the ROC curve, an optimal cut-point value of 163 U/l was obtained to differentiate between patients with longer and shorter survival. The Kaplan-Meier curves, showed as poor prognostic factors LDHb serum levels higher than 163 U/l, linked to a survival below 60 months. Ryan et al. (6), using the data from patients treated with AA+P extracted from COU-AA-302, increase this threshold to levels above 234 U/l.

The role that the AP serum biomarker plays, which in bladder and gastric cancers is indicative of an independent bone metastasis risk factor, is not as clear in PC cases. It is thought that at the start of metastatic disease, AP marks the bone change (22). A recent meta-analytic study, based on data from 63 studies, reports that high AP at baseline is associated with poor overall survival rates and PFS in patients suffering from PC (23). Likewise, a series of studies have used baseline levels and AP changes during treatment with AA+P as tools to identify patient groups with different risks of bone lesion progression (2, 13, 22, 23). Thus, baseline levels of AP≤119 U/l were linked to longer treatment time and a longer span until the disease progression to mCRPC (2). Similarly, a recent study that accesses the data of 546 patients from the COU-AA-302 study, points out that an increase in AP values above 20% during treatment with (AA+P) is related to bone lesion progression (13). The present study did not find significant differences between patients with longer and shorter survival when using AP base line levels; yet it was a prognostic marker of the AP serum levels at six months, resulting in an optimal cut-point value of 56 U/l. The Kaplan-Meier curve showed a very low survival rate after 60 months (14.3%) in patients with levels of AP 6 m >56 U/l.

The PSA response after having initiated AA+P treatment is a positive prognostic factor if data from studies such as COU-AA-302 and 301 is taken into account, where a correlation between the overall survival rate and PSA kinetics was established (24). In contrast, the lack of PSA response could be considered as a potential biomarker for patient selection - it could be eliminated and substituted by an alternative or an additional marker. Within the present study, 16 patients exhibited a sharp increase in PSA levels at the start of treatment (known as “PSA flare”), yet no significant differences were found in their clinical results. Although the Prostate Cancer Clinical Trials Working Group (PCWG2) (25) group criteria suggest that early changes in PSA (before the 12th week) should not be taken into consideration when determining the response to AA+P treatment, different studies have shown a strong link between early PSA level reduction and overall survival rates and rPFS (26, 27). Consequently, and given the frequent occurrence of this phenomenon as well as the inconsistent data, it is recommended that PSA levels are interpreted from week 12, and that early interruption of those patients’ treatment should be avoided (27). However, this decision should be revisited when no biochemical response beyond week 12 is observed. Poon et al. (11) demonstrated that there was no survival improvement in patients with no response to PSA after 12 weeks and when treatment was maintained.

A meticulous interpretation of PSA levels from 12 weeks onwards is suggested, to evaluate whether treatment should be suspended in patients for whom no clinical benefit, even in imaging tests (carried out during this 12th week), can be observed.

In this series of patients, PSA levels at six months showed higher precision in differentiating between patients with longer and shorter survival, with an optimal cut-point value of 0.95 ng/ml. Conversely, significant differences in the Kaplan-Meier curve were not found, so survival at 60 months of treatment could not be predicted. This could be due to the small sample size of the present study.

In turn, a 30%, 50% and 90% decrease in PSA levels, as well as measurements connected to PSA kinetics, are associated with survival and radiological progression in different studies (26). In the present study, a 90% PSA reduction was linked to a higher survival rate, although once again survival at 60 months of treatment could not be predicted.

In summary, and after having interpreted the results gathered within this series of patients, the following prognostic markers can be defined: 1) Baseline LDH and AP at six months are considered as markers with a highly demonstrated prognostic value: significant differences between patients with longer and shorter survival were exhibited, as well as statistical significance shown in the Kaplan-Meier curve, resulting in an estimation of survival at 60 months. 2) PSA levels at six months is a possible prognostic marker: significant differences between patients with higher and lower survival rates were exhibited, but not reflected in the Kaplan-Meier curve.

These serum markers can be prognostic factors of the evolution and follow-up of patients with mCRPC, considering that there are different methods to determine them, and that differences can be found when quantifying them.

This work identified prognostic factors in patients with mCRPC treated with AA+P, studying a large number of variables, analysing the differences between patients with longer and shorter survival rates, applying ROC curves analysis and Kaplan-Meier survival curves. One of the observed limitations is the small number of studied patients (n=53), which was insufficient to carry out a multi-variable analysis and to develop a probability model. Studies with a higher number of patients are necessary, as they allow for multi-variable analysis, and can confirm the prognostic factors identified in the present study.

The retrospective design of this study adds certain limitations. The survival in patients who were still alive after having gathered all the data was the survival at said point (which was always above 24 months, as it was the minimum follow-up period established so that they could be included in the higher survival group of this study), yet there was no way of knowing the exact survival age that these patients could reach. Nonetheless, the survival rate of the deceased patients during the study was precise.