Imaging and biochemical predictive biomarkers of overall survival and treatment response in patients receiving 223Ra-dichloride treatment in clinical practice

Objectives The aim of the study was to investigate whether the extent of bone disease as demonstrated on Tc-99m-MDP bone scintigraphy and biochemical response as measured by alkaline phosphatase (ALP) intra-treatment change may predict the outcome and overall survival in patients receiving 223Ra-dicholoride treatments. We also aim to present initial institutional experience with 223Ra-dichloride in patients with castrate resistant prostate cancer in clinical practice.

Methods All patients treated in period March 2014 to December 2015 were included. The extent of bone disease (EOBD) was classified as 1 (less than 6 metastases), 2 (6-20), 3 (>20), 4 (superscan). Standard biochemistry (FBC, PSA, ALP, LDH and SA), pain flare and serious skeletal related events (SSRE) were recorded in all patients. Baseline ALP, % change between the baseline ALP and all subsequent treatments and the greatest % reduction (ALP response max) were calculated during the entire course of 223Ra-treatment and correlated with survival and latest follow-up. Overall survival was defined as the time from first clinic visit to the date of death, regardless of cause.

Results Fifty-two patients commenced 223Ra-dichloride [7/52 (13%) EOBD1, 16/52 (31%) EOBD2, 15/52 (29%) EOBD3, 14/52 (27%) EOBD4] were treated by 182 cycles. No patient in EOBD1 died, with median follow-up of 170 days and an interquartile range (IQR) of 104-216 days. In EOBD2 6/16 (31%), EOBD3 8/15 (53%), EOBD4 11/14 (79%) died. Median survivals and IQRs for each group were 225 days (IQR 218-249), 170 days (IQR 131-235), 104 days (IQR 58-194), respectively. A log-rank test was applied to Kaplan-Meier survival analysis to determine that a statistically significant difference existed between EOBD2/3 and EOBD4 (p < 0.03 and p < 0.05, respectively), but there was no significant difference between EOBD2 and EOBD3. Total ALP decreased during the treatment with median % ALP reduction compared to baseline after 1st, 2nd, 3rd, 4th, 5th, 6th subsequent treatments were 29%, 38%, 44%, 48%, 56% and 51%, respectively. However, the baseline ALP did not predict survival. The greatest ALP % reduction (ALP response max) was usually patient specific and classified according to ranges of ALP response into 2 groups: responders (R), with 3 subgroups: R1 70-100%, R2 40-69%, R3 10-39% and non-responders (NR) with increase or <10% reduction from baseline. Patients with normal baseline ALP (13/52, 25%) were excluded. A strong correlation was observed (R^2 = 0.96) between the greatest % ALP reduction (response max) and the longest median survival. A log-rank test of the survival analysis was performed on seven biochemical ALP non-responders (NR) compared to responders (R). This showed that NRs had statistically significantly shorter survival times (p < 0.01) of 109 days (IQR 60-90) compared to 181 days (IQR 100-310) for biochemical responders. Thirteen out of 52 (25%) patients experienced pain flare. Three of these were severe, 1 experienced a moderate pain flare response, while the remainder of these patients experienced a mild pain flare response. SSEs were recorded in 5/52 (9.6%) patients.

Conclusions Extent of bone disease as demonstrated by bone scintigraphy is a predictive imaging biomarker of the overall survival of patients treated with 223Ra-dicholoride treatments. ALP response (max) and no ALP response in relation to the baseline ALP are likely to be independent biochemical predictors of outcome in clinical practice, which may help tailoring this molecular radiotherapy to individual patient profiles with precision.