In this study, we assessed the usefulness of early PSA change after ADT administration as a predictive marker for mHSPC patients. We found that %PSA ≥ 1 was a strong predictive factor of OS. In addition, Gleason score ≥ 8, EOD ≥ 2, and %PSA ≥ 1 were significantly associated with shorter time to CRPC. For diagnostic performance, the AUC of CRPC diagnosis for differentiating the high-risk group from the other group was feasible (0.822).
There have been several reports focusing on PSA-related variables, including initial PSA levels and PSA kinetics, which are the most frequently assessed biomarkers in mHSPC . Among the PSA kinetic variables, nadir PSA and time to the nadir PSA are promising biomarkers for mHSPC [17, 18]. However, previous studies reported that the median time to the nadir PSA was 6–10 months [17, 18], which means that the prognosis was predictable only after more than half a year following the initiation of ADT. On the other hand, in our report, the prognosis can be predicted at 3 months, and the judgment can be made in a short period. Ji et al. reported that a decline in PSA levels of > 11 ng/mL per month after initial ADT was significantly associated with an increased risk of progression to CRPC . Therefore, to predict the outcomes of mHSPC patients, 3-month %PSA might be a simple and convenient biomarker for the prediction of clinical outcomes. Sato et al.  also reported that a group whose PSA level decreased by 98.5% at 12 weeks after the initiation of ADT had significantly increased progression-free survival and OS. In our study, 3-month %PSA ≥ 1 was an independent predictive marker, consistent with Sato’s results. In their report, patients with visceral metastases were relatively low at 3.3%. On the other hand, in our report, 24% of patients had visceral metastases, which have relatively poor prognoses. It is interesting to note that, even in such patients, an early decline in PSA levels contributed to significantly longer OS. It is suggested that the 3-month PSA < 1% group may have a better prognosis in mHSPC.
In previous studies, serum bone markers , circulating tumor cells (CTC),  and single nucleotide polymorphisms  were identified as potential prognostic markers for patients with mHSPC. However, they are difficult and very expensive to adopt in clinical practice. On the other hand, the early changes in PSA are simple and inexpensive to evaluate and are clinically useful.
The independent factors affecting the time to CRPC in our study were: 1) Gleason score ≥ 8, 2) EOD ≥ 2, and 3) %PSA ≥ 1. Two factors (Gleason score ≥ 8 and EOD ≥ 2), except for visceral metastasis, are compatible with the results of the initial trial of abiraterone acetate for high-risk mHSPC in Japan. In the LATITUDE trial , visceral metastasis was listed as an item of high risk; however, in our study, it was not an independent predictor of poor prognosis. Distant metastasis, particularly visceral metastasis, is an important negative prognostic factor in prostate cancers [3, 24–27]. Cui et al. reported that OS of prostate cancer patients with visceral metastasis with lung metastasis had a better prognosis than brain or liver metastasis . In our study, 11 out of 17 patients (65%) with visceral metastasis had lung-only metastasis. This might be reason that visceral metastasis was not shown to be a high-risk factor.
The present study reported that the group with 3-month %PSA ≥ 1 had poor prognoses, which suggests that it is a high-risk factor in mHSPC. In bone-metastatic prostate cancer patients, it is reported that age, T stage, PSA, Gleason score and EOD are useful as prognostic nomogram factors . In our study, the AUC of the time to CRPC for differentiating the high-risk group that satisfied our three independent factors (Gleason score ≥ 8, EOD ≥ 2, and 3-month %PSA ≥ 1) was high at 0.822. Thus, our cut off level, 3-month %PSA ≥ 1, is feasible and might be useful for a new scoring method for detecting high-risk mHSPC.