The Development and Validation of a Prognostic Nomogram and Nomogram Application Software Among Men Treated with Abiraterone Acetate and/or Enzalutamide for Metastatic Castration-resistant Prostate Cancer

Background: With widespread medication choices for metastatic castration-resistant prostate cancer (mCRPC) is now available, on the other hand biomarker to predict the ecacy of each mCRPC treatment has not been established. Objective: This study developed prognostic nomogram to predict prognosis in CRPC patients who received abiraterone acetate (ABI) and/or enzalutamide (ENZ). Design, Setting, and Participants: A total of 568 mCRPC patients received ABI and/or ENZ from 2012 to 2017 were enrolled in this study. We developed prognostic nomogram based on the risk factors by Cox proportional hazards regression model.


Introduction
A recent clinical trial revealed the e cacy of abiraterone acetate (ABI), enzalutamide (ENZ), Radium-223 (Ra-223), and cabazitaxel in addition to docetaxel chemotherapy in metastatic castration-resistant prostate cancer (mCPRC) patients [1][2][3]. In the next few years, poly(ADP-ribose) polymerase (PARP) inhibitors and immune-checkpoint inhibitors are expected to be used in clinical practice to similar ends [4,5]. With widespread medication choices now available, clinicians should take care to select the most appropriate medicine in order not to lose their chance to administer the best therapy possible to a patient. Predicting the prognosis is thus important, because a lack of biomarker to predict the e cacy of each mCRPC treatment.
Recent studies have demonstrated the e cacy of tumor markers for predicting the prognosis, such as in ammatory markers, including the neutrophil-to-lymphocyte ratio, monocyte-to-lymphocyte ration, and platelet-to-lymphocyte ratio, as well as the alkaline phosphatase (ALP) and lactate dehydrogenase (LDH) levels [6]. However, to more accurately predict the prognosis, a nomogram using multiple prognostic parameters is needed [7,8].
In the present study, we developed and validated a nomogram for predicting the prognosis of mCRPC patients who receive ABI and/or ENZ treatment using clinical information and also developed application software (app) to facilitate clinical usage.

Patients
A total of 568 metastatic CRPC (mCRPC) patients received ABI and/or ENZ in Yokohama City University, Nagoya University, Kitasato University, and a liated hospitals from 2012 to 2017. All cases were pathologically con rmed to have prostate cancer and received androgen deprivation therapy (ADT) but proved refractory.
The institutional review board of Yokohama City University Medical Center was approved this study (D1603004). The de nition of CRPC was set by the Prostate Cancer Working Group 2 [9]. Patients' background characteristics, including the initial prostate-speci c antigen (PSA) level, initial metastatic status, Gleason score, observation period, time to CRPC, pre-chemotherapy, age at baseline, PSA at baseline, ALP at baseline, LDH as baseline, and initial ABI/ENZ treatment, are shown in Table 1.

Nomogram development
The nomogram for the OS was developed using a Cox proportional hazards regression model with the age, initial PSA, initial stage (M0/M1), Gleason score, time to CRPC, Chemotherapy, PSA at baseline, ALP at baseline, and LDH at baseline as the predictors. Calibration was performed using the methods described by Iasonos et al. [10]. The data were randomly separated into training and validation data sets to calibrate the nomogram prediction. The prediction was evaluated by comparing the predicted survival probability at two years with the observed survival probability using the training and validation data sets. The nomogram was also assessed for discriminatory ability with the concordance index (C-index). We repeated 5-fold cross-validation 2000 times to estimate the C-index and reported the means of the estimated C-index using the training and validation sets.

Results
The median (range) observational period was 13.   and four clinically important risk factors (age, initial PSA, initial metastatic status, and Gleason score). Figures 3a and 3b show the calibrations of the nomogram for the two-year survival. The blue diagonal line indicated the ideal reference line at which predicted the probabilities match the observed proportions. The vertical lines across the blue line represent the nomogram-predicted probabilities grouped for each of the four quartile groups, along with the respective 95% con dence intervals. From Fig. 3a, we can see the predicted survival rate from the nomogram was well correlated with the actual observation of the twoyear survival in the training data set. Figure 3b shows the calibration of the nomogram for the two-year survival using randomly selected validation data set. We can also see that the predicted survival rate from the nomogram was well-correlated with the actual observation of the two-year survival in the validation set. The means of estimated c-index were 0.72 for training data sets and 0.71 for validation data sets in the 5-fold cross-validations.
We also developed app to easily use these ndings in daily clinical practice [ Fig. 4] The app were developed for the Android and iOS systems [ Fig. 5].

Discussion
This study developed and validated a nomogram for predicting the prognosis of mCRPC patients who receive ABI and/or ENZ. This nomogram used the initial PSA, initial metastasis status, Gleason score, time to CRPC, previous use of docetaxel or not, age at ABI/ENZ installation, and laboratory data, including the PSA/ALP/LDH at the time of ABI/ENZ installation. The prognosis of mCRPC varies among metastatic lesions. Regarding non-metastatic CRPC (m0CRPC), the PROSPER, SPARTAN, and ARAMIS studies showed the median radiographic progression-free survival (rPFS) to be around 36.6 to 40.4 months [11][12][13]. Although the nal OS was not reached, the OS was expected to be around 67.0 to 73.9 in both groups [14]. Regarding metastatic pre-docetaxel chemotherapy CRPC, the PREVAIL or COU-AA-302 studies showed that the OS was around 32.4 to 34.7 months [1,15]. Furthermore, regarding metastatic postdocetaxel chemotherapy CRPC, the AFFIRM and COUA-AA-301 showed that the OS was around 14.8 to 18.4 months [2,16]. Finally, in real-world metastatic CRPC, the OS was found to be 31.6 months in cases of lymph-node metastasis, 21.3 months in cases of bone metastasis, 19.4 months in cases of lung metastasis, and 13.5 months in cases of liver metastasis [17]. While the OS was speculated in metastatic site, the detailed prognostic estimation using multiple risk factors has not been established. To make right treatment decisions for the right patient at right timing, a detailed prognosis estimation is needed.
Recent studies have shown that elevated tumor markers, such as LDH and ALP, and some in ammatory markers are poor prognostic factors for CRPC [6]. Zhao et al. developed a prognostic nomogram for CRPC among Chinese patients using a 449-patient cohort [18]. This nomogram used the Gleason score, presence of intraductal carcinoma of the prostate, baseline ALP/PSA/Hb, and the Eastern Cooperative Oncology Group performance status. Yang et al. also developed a nomogram using the presence of liver metastasis, hemoglobin level, and time from initial ADT to ABI in 110 Chinese CRPC patients [19]. Lin et al. developed a nomogram using the PSA-doubling time, time to PSA progression, and presence of pain in 167 Chinese CRPC patients [20]. This is the rst study to predict the prognosis in mCRPC patients in a total of 589 patients who received ABI/ENZ.
In mCRPC treatment, newly established medications are more expensive than ADT or docetaxel treatment, which eventually affects a country's insurance system [6,21]. Previous studies also created nomograms to predict the prognosis in Chinese CRPC patients [ref]. The economic range in Asia is vast, so nomograms speci c to each country are needed. Under the Japanese medical insurance system, all patients received government-approved CRPC medication, including ABI/ENZ, Ra-223, docetaxel, and cabazitaxel. Speleucel T is not approved in Japan. The present ndings are expected to bene t Japanese mCRPC patients who are introduced to ABI/ENZ.
Several limitations associated with the present study warrant mention. First, this study used a retrospective cohort divided into a training group and control group from Japanese multicenter hospitals.
Most of the hospitals were third referral cancer centers. Thus, a further study is needed to con rm the accuracy of these ndings using all hospitals, including private clinics. Second, other mCRPC treatment drugs, such as PARP inhibitors and immune-checkpoint inhibitors, will be available in the near future. Once these new treatments are approved, additional validation will likely be needed.
In conclusion, we developed and validated a nomogram for predicting the prognosis of mCRPC patients who receive ABI/ENZ treatment using clinical information and also developed app to facilitate clinical usage.

Conclusion
We developed nomograms to predict the OS for Japanese mCRPC patients who received ABI and/or ENZ. This nomogram might help clinicians make treatment decisions.

Declarations
Competing interests Figure 1 Kaplan-Meier curve for the overall survival in metastatic castration-resistant prostate cancer patients treated with abiraterone and/or enzalutamide Nomogram for predicting the one-and two-year survival among men treated with abiraterone acetate and/or enzalutamide for metastatic castration-resistant prostate cancer.
Page 14/15   Android and iphone/iOS app to predict prognosis for 1-year and 2-year survival among men treated with abiraterone acetate and/or enzalutamide for metastatic castration-resistant prostate cancer.

Figure 5
QR code to download the a) Android and b) iPhone/iPad app.