Prognostic Value of Posttherapy SPECT/CT for Overall Survival in Patients Undergoing [¹⁷⁷Lu]Lu-PSMA-617 Radiopharmaceutical Therapy: Results from 3 Clinical Trials

Visual Abstract

Abstract

Data are emerging on the prognostic significance of quantitative changes on posttherapy SPECT/CT in patients with metastatic castration-resistant prostate cancer (mCRPC) receiving [¹⁷⁷Lu]Lu-PSMA-617. Our objective was to assess quantitative and visual changes on posttherapy SPECT/CT as prognostic biomarkers for overall survival (OS) among patients in 3 clinical trials: LuPSMA Phase 2 (ANZCTR12615000912583), LuPARP (NCT03874884), and PRINCE (NCT03658447)]. Methods: We segmented the total tumor burden on posttherapy [¹⁷⁷Lu]Lu-PSMA-617 SPECT/CT using an SUV threshold of 3 to measure SUV_max, SUV_mean, metabolic tumor volume (MTV), and total lesion activity (TLA). We assessed the prognostic value of changes in these quantitative parameters and new lesions identified visually on SPECT/CT after cycle 2 for OS using the Cox proportional hazards model, with age, Gleason score, and change in prostate-specific antigen (PSA) as covariates. Results: Eighty-five patients with mCRPC were analyzed (46 from LuPSMA Phase 2, 25 from PRINCE, and 14 from LuPARP). Patients eligible for inclusion had received at least 2 cycles of [¹⁷⁷Lu]Lu-PSMA-617 with a follow-up time of at least 12 mo. Among these patients, 18 (21.2%) had new metastases visible on cycle 2 SPECT/CT, and this was prognostic for OS in univariate (hazard ratio [HR], 2.38; 95% CI, 1.36–4.18; P = 0.002) and multivariate (HR, 2.85; 95% CI, 1.36–5.98; P = 0.01) analyses. Seven (8.2%) patients with PSA reductions had new lesions on posttherapy SPECT/CT. Reductions in TLA (HR, 0.98; 95% CI, 0.97–1.00; P = 0.016) and MTV (HR, 0.98; 95% CI, 0.96–1.00; P = 0.009) (per 10% increase for both) were associated with OS on univariate analysis but not on multivariate analysis. Changes in SUV_max and SUV_mean were not associated with OS. There was moderate correlation among changes in PSA from cycle 1 to cycle 2 and MTV (correlation coefficient = 0.55; 95% CI, 0.39–0.69; P < 0.001) and TLA (correlation coefficient = 0.56; 95% CI, 0.40–0.69; P < 0.001). Conclusion: The presence of new metastases on posttherapy SPECT/CT after cycle 2 is an independent prognostic biomarker for OS in patients with mCRPC and could guide future prospective research to improve treatment strategies for patients with poor prognoses.

Radiopharmaceutical therapy with [¹⁷⁷Lu]Lu-PSMA-617 recently received approval from the U.S. Food and Drug Administration for the treatment of metastatic castration-resistant prostate cancer (mCRPC), as its use is associated with an improvement in overall survival (OS) (1,2). While this treatment has been widely adopted globally, the clinical utility of posttreatment SPECT/CT imaging is unclear. Although many centers routinely use SPECT/CT, the VISION trial did not include posttreatment imaging (1). In contrast, the TheraP trial mandated posttreatment SPECT/CT imaging to determine whether to pause treatment in a subset of patients showing an exceptional response (3).

The role of posttherapy imaging is continually evolving. In addition to providing quality assurance for radiopharmaceutical localization, SPECT/CT provides valuable clinical information on temporospatial changes in tumor sites and volume, enabling real-time response assessment. Changes in prostate-specific antigen (PSA) levels are prognostic markers in patients treated with [¹⁷⁷Lu]Lu-PSMA-617 therapy (4). Quantitative changes on posttherapy SPECT/CT have been demonstrated as prognostic for progression-free survival (PFS) but not OS in the LuPIN trial and a [¹⁷⁷Lu]Lu-PSMA-I&T registry (5,6). We sought to evaluate posttherapy SPECT/CT finding as prognostic biomarkers for OS in patients with mCRPC receiving [¹⁷⁷Lu]Lu-PSMA-617.

MATERIALS AND METHODS

Design, Study Population, and Objectives

We conducted a retrospective analysis of the posttherapy SPECT/CT scans from patients with mCRPC who were treated with [¹⁷⁷Lu]Lu-PSMA-617 in the LuPSMA (ANZCTR12615000912583) (7), LuPARP (NCT03874884) (8), and PRINCE (NCT03658447) (9) clinical trials at the Peter MacCallum Cancer Centre. Patients who had undergone SPECT/CT after treatment with [¹⁷⁷Lu]Lu-PSMA-617 and at least12 mo of follow-up (or documented progression or death prior) were eligible for inclusion. We excluded patients who had fewer than 2 cycles of [¹⁷⁷Lu]Lu-PSMA-617, as imaging at a single time point would not allow for comparisons to be made. This retrospective study of prospectively collected data was approved by the institutional ethics review board (PMC92091). The requirement for separate informed consent was waived by the ethics board, as patients had already consented for future research as part of each study. All 3 studies were investigator-initiated and institution-sponsored.

We aimed to evaluate the prognostic value of 2 factors—changes in quantitative parameters and the appearance of new PSMA-avid metastases on SPECT/CT between cycles 1 and 2—on PFS and OS. Additionally, we sought to correlate the changes in quantitative parameters on SPECT/CT between cycles 1 and 2 with changes in PSA measured before starting [¹⁷⁷Lu]Lu-PSMA-617 and 5–6 wk after cycle 2 of [¹⁷⁷Lu]Lu-PSMA-617. The impact of imaging on clinical management also was assessed.

Procedures

Serum PSA values were measured within 6 wk of initiation of treatment and at 3 and 5 wk after each cycle of treatment. The most recent PSA value before each [¹⁷⁷Lu]Lu-PSMA-617 cycle was utilized to assess response.

Quantitative SPECT/CT scans were acquired from vertex to mid-thigh approximately 24 h after intravenous administration of [¹⁷⁷Lu]Lu-PSMA-617 (Intevo 16; Siemens Healthineers). The serial posttherapy SPECT/CT images were reviewed by 2 experienced nuclear medicine physicians, who performed total tumor burden segmentation with a semiautomated workflow using an SUV threshold of 3 (MIM version 7.2; MIM Software). Normal structures, such as the salivary glands, kidneys, urinary bladder, and bowel, were excluded. The quantitative parameters measured on SPECT/CT included SUV_max, SUV_mean, metabolic tumor volume (MTV), and total lesion activity (TLA), which is the product of MTV and SUV_mean. The reviewers also analyzed the posttherapy SPECT/CT images for the appearance of new PSMA-avid metastases with uptake greater than that of the liver after cycle 2. We did not require a CT correlate for bone lesions to be classified as metastases.

Clinical notes were reviewed to evaluate the impact of posttherapy SPECT/CT on the overall management of the patient. Disease progression and decisions on pausing treatment in the trials were determined based on criteria published in the respective trial protocols (7–9). Posttherapy SPECT/CT was defined as having a high impact on management when a clinical decision was made, based on the images, to change therapy to a different line, add radiotherapy to treat metastases, pause [¹⁷⁷Lu]Lu-PSMA-617 for exceptional response, or discontinue [¹⁷⁷Lu]Lu-PSMA-617 because of the appearance of new lesions (i.e., unequivocal progression).

Statistical Methods

All statistical analyses were performed in R version 4.3.1. OS was defined as the time from cycle 2 administration to the date of death from any cause. PFS was defined as the time from cycle 2 administration to the date of biochemical progression (using Prostate Cancer Working Group 3 criteria) or death from any cause.

The correlation between change in quantitative parameters on SPECT and change in PSA level was performed using Spearman correlation. Kaplan–Meier curves were used to estimate PFS and OS. The prognostic value of the change in quantitative parameters and the appearance of new lesions on SPECT/CT after cycle 2 on OS and PFS was assessed using the Cox proportional hazards model. For each of these variables of interest, a Cox model was fitted, with age, Gleason score, and change in PSA level from cycle 1 to cycle 2 as covariates (i.e., 1 Cox model for each SPECT parameter). Additionally, SUV_mean, SUV_max, MTV, and TLA were modeled with splines to assess linearity. The assumption of proportional hazards was examined. A time-dependent coefficient Cox model was used when the assumption of proportional hazards was not satisfied. The impact of SPECT/CT on clinical management was described as count and percentage for each category.

RESULTS

From October 2016 to June 2021, a total of 89 patients with mCRPC treated with [¹⁷⁷Lu]Lu-PSMA-617 were screened for study inclusion, including all 50 patients from the LuPSMA Phase 2 trial, 25 patients from the PRINCE trial, and the first 14 patients who received treatment in the LuPARP trial. Patients were eligible if they had a follow-up time of at least 12 mo at the time of analysis. Four patients were excluded from the analysis: 1 patient from the LuPSMA Phase 2 trial because of technical issues resulting in unreliable quantitative parameters on SPECT/CT and 3 patients who received only 1 cycle of [¹⁷⁷Lu]Lu-PSMA-617 because of early progression or death. The remaining 85 patients were included in the analysis, with baseline characteristics summarized in Table 1.

TABLE 1.

Patient Characteristics at Baseline (n = 85)

Characteristic	Value
Age (y)
Mean ± SD	70.7 ± 7.3
Median and range	71 (50–86)
Interquartile range	67–75
Gleason score
6	1 (1)
7	29 (38)
8	8 (10)
9	34 (44)
10	5 (6)
NA	8
Patient selection
LuPSMA trial	46
PRINCE trial	25
LuPARP trial	14
Baseline PSA level (µg/L)	366.2 (31.1–289.4)
Prior treatment
Abiraterone/enzalutamide	85
Prior chemotherapy (taxane)	76
Median follow-up time (mo)	33.5
No. [177Lu]Lu-PSMA-617 cycles
2	11 (13)
3	14 (16)
4	36 (42)
5	9 (11)
6	15 (18)

NA = not available.

Qualitative data expressed as number, followed by percentage in parentheses; continuous data expressed as median, followed by interquartile range in parentheses.

There was a moderate correlation between the change in both MTV and TLA with the percent change in PSA values from cycle 1 (correlation coefficient, 0.55; 95% CI, 0.39–0.69; P < 0.001) to cycle 2 (correlation coefficient, 0.56; 95% CI, 0.40–0.69; P < 0.001) (Fig. 1).

FIGURE 1.

Correlation between change in TLA and PSA (A) and MTV and PSA (B). C1 = cycle 1; C2 = cycle 2.

Survival Analysis and Prognostic Factors

For the entire cohort, the median OS was 14.3 mo (95% CI, 10.9–16.9 mo), and the median PFS was 5.7 mo (95% CI, 4.1–6.9 mo). The median follow-up period was 33.5 mo (Fig. 2).

FIGURE 2.

OS (A) and PFS (B) of patient cohort based on appearance of new PSMA-avid metastases on cycle 2 SPECT imaging.

A reduction in TLA and MTV between cycles 1 and 2 was prognostic for OS on univariate analysis. However, this was not the case on multivariate analysis when adjusting for age, Gleason score, and reduction in PSA (Table 2). Changes in SUV_max and SUV_mean were not prognostic for OS.

TABLE 2.

Cox Regression of SPECT Quantitative Parameters for OS

	Univariate analysis		Adjusted by age, Gleason score, and reduction in PSA
Variable	HR	P	HR	P
Reduction in SUVmax*	0.97 (0.93–1.02)	0.203	0.98 (0.93–1.03)	0.383
Reduction in SUVmean*	0.94 (0.85–1.05)	0.269	0.98 (0.86–1.11)	0.706
Reduction in TLA*	0.98 (0.97–1.00)	0.016	0.99 (0.97–1.01)	0.242
Reduction in MTV*	0.98 (0.96–1.00)	0.009	0.98 (0.95–1.01)	0.186
New lesion on SPECT after cycle 2†	2.38 (1.36–4.18)	0.002	2.85 (1.36–5.98)	0.010

^*Per 10% increase in reduction percentage; 71 events in 85 patients.

^†New lesions found in 17 events in 18 patients. No new lesions found in 54 events in 67 patients.

Continuous data are 95% CI.

The appearance of new lesions on cycle 2 posttherapy SPECT/CT was prognostic for OS on both univariate (hazard ratio [HR], 2.38; 95% CI, 1.36–4.18; P = 0.002) and multivariate (HR, 2.85; 95% CI, 1.36–5.98; P = 0.01) analyses. The median OS for patients with new lesions was 9.2 mo (95% CI, 4.4–11.8 mo) compared with 15.6 mo (95% CI, 11.5–19.9 mo) for those without new lesions. Notably, 8% of patients (7/85) with PSA reduction after cycle 2 had new lesions detected on posttherapy SPECT/CT (Fig. 3).

FIGURE 3.

Waterfall plot showing change in PSA from cycle 1 to cycle 2 (A) and MTV change from cycle 1 to cycle 2 (B).

For PFS, a Cox proportional hazards model was fit for each PET parameter, adjusting for age, Gleason score, and reduction in PSA from cycle 1 to cycle 2. Although there was no clear evidence of nonlinearity, there was a suggestion of a nonproportional hazard for new lesions on PFS (P = 0.013 and visual inspection of residuals). The residual plot suggested that a time-dependent coefficient using 2 mo as the cutoff may be appropriate and fulfill model assumption. The appearance of new lesions was prognostic for PFS within 2 mo of cycle 2 (HR, 14.9; 95% CI, 4.80–46.40; P < 0.001) but not beyond 2 mo (HR, 1.13; 95% CI, 0.35–3.70; P = 0.836), suggesting that a new lesion found with SPECT is a potential marker of immediate PSA progression.

There was consideration to model new lesions on SPECT from each cycle as a time-varying covariate. However, after inspecting new lesions in cycle 3, almost all patients (n = 8) with new lesions in cycle 3 progressed at or around the SPECT after cycle 3. Therefore, it appears that a new lesion found on SPECT at later time points does not serve as a predictive marker of future progression but rather a surrogate for the current state of the disease.

Impact on Management

Posttherapy SPECT/CT was observed to have a high impact on clinical management in 20 (24%) of 85 patients. Among these, [¹⁷⁷Lu]Lu-PSMA-617 was paused because of exceptional response in 9 patients (11%) (Fig. 4), [¹⁷⁷Lu]Lu-PSMA-617 was discontinued for unequivocal progression in 5 patients (6%) (Fig. 5), and external-beam radiotherapy was recommended for and administered to 6 patients (7%). The appearance of new lesions, which was deemed unequivocal progression, preceded increases in PSA levels in all 5 patients for whom [¹⁷⁷Lu]Lu-PSMA-617 was discontinued (Fig. 6).

FIGURE 4.

Posttherapy SPECT images demonstrating favorable response resulting in pausing treatment after cycle 3. PSA dropped from 2.1 µg/L before cycle 1 to 0.6 µg/L after cycle 3.

FIGURE 5.

Posttherapy SPECT images of patient showing progressive metastases through cycles 1–3 of [¹⁷⁷Lu]Lu-PSMA. Baseline PSA was 486 µg/L and increased to 855 µg/L before cycle 2 and 1,347 μg/L after cycle 2. Bone scan and CT criteria for progression on clinical trial were met at date later than cycle 3 administration.

FIGURE 6.

Posttherapy SPECT images of patient with new PSMA-avid metastases after cycle 2. PSA dropped from 69 µg/L before treatment to 11 µg/L 5 wk after cycle 1. New lesions (circled) were noticed on posttherapy SPECT after cycle 2. PSA progression to 35 µg/L from nadir was noticed only 5 wk after cycle 3. On cycle 4 posttherapy images, there is marked progression of PSMA-avid metastases.

DISCUSSION

Although there are strong data on baseline predictive and prognostic biomarkers, including quantitative PET parameters, for outcomes of [¹⁷⁷Lu]Lu-PSMA-617 therapy (10–15), data are limited on the significance of posttherapy imaging and parameters derived therewith. A decline in PSA values after cycle 1 is known to have a positive impact on outcomes, whereas an increased Gleason score and nodal and visceral metastases, along with increased alkaline phosphatase levels and the use of analgesics, negatively affect outcomes (16,17).

Posttherapy SPECT/CT imaging is an established practice in many nuclear medicine centers but varies by country. It is an inexpensive scan that can be performed at most centers and utilizes the natural γ-emissions from [¹⁷⁷Lu]Lu-PSMA-617 for visualization. Importantly, it enables visualization of distribution of the radiotracer, acting as a marker of radiation delivery, and aids in response assessment through the course of treatment. Although traditionally performed with whole-body planar imaging, SPECT imaging that can be fused with CT images for anatomic localization enables estimation of quantitative parameters, such as SUV at lesion sites (18) and tumor burden, with the use of image-processing techniques. At 24 h after [¹⁷⁷Lu]Lu-PSMA-617 therapy, there is high tumor and low background activity, enabling workflows to readily estimate MTV or TLA (MTV × SUV_mean). We aimed to evaluate the prognostic significance of these SPECT-derived parameters and utility of the SPECT/CT imaging on clinical decision-making. Early (4 h) and delayed (72 h) images were also acquired for a few patients in the LuPSMA Phase 2 trial but not in the other 2 trials. However, we limited our analysis to 24-h images only. We used an SUV threshold of 3 for tumor delineation and found it useful, with a low incidence of false positives. We have found that, at 24 h after injection, there was high retention in tumors and that benign conditions, such as incidental benign bone lesions, showed washout, which is less than an SUV of 3. This is unlike PSMA PET, which is acquired 1 h after injection, where false positives can be noted.

In this study, we evaluated images obtained after cycle 1 and cycle 2, with a 6-wk interval between them. Although changes in SUV_max, SUV_mean, MTV, and TLA were prognostic for OS, these did not provide additional prognostic value to clinical parameters. A previous study by John et al. (6) used similar time points and demonstrated consistent results. On the other hand, Pathmanandavel et al. (5) evaluated the role of posttherapy imaging after cycles 1 and 3 (at 12 wk) and found that an increase in total tumor volume at 12 wk was prognostic for outcomes. Another study also showed that a change in total tumor volume was prognostic for OS, whereas changes in SUV_max and SUV_mean were not prognostic for both PFS or OS (19).

In our study, the strongest prognostic parameter was the detection of new metastases after cycle 2, surpassing the value of quantitative biomarkers and being superior to clinical parameters. This adds value beyond PSA or quantitative measures, as some patients develop new metastases despite decreasing PSA values. Moreover, visual analysis is rapid and does not require specialized software for image processing. In a recent study, Demirci et al. (20) also found that poor OS was associated with the appearance of new lesions at the start of cycles 2 and 3, along with baseline tumor burden. These factors were independently prognostic of changes in PSA levels. Similar to our study, their study also demonstrated a correlation between the change in total tumor volume and change in PSA levels. We did not analyze other potential markers that have been studied, such as distance between lesions (21) or size of the largest skeletal lesion (22).

Posttherapy imaging influenced clinical management in up to 24% of the patients. These posttherapy SPECT images provided guidance to pause treatment because of an exceptional response, where there was loss of PSMA-target expression; direct focal treatment for oligoprogressive sites; and discontinue therapy because of the appearance of multiple new lesions, indicative of progressive disease. Furthermore, a recent study by Yadav et al. (23) demonstrated change in management based on posttherapy imaging in up to 49% of patients.

This study advances the field because, unlike RECIP (24), PET tumor to salivary gland score (25), and heterogeneity and intensity of tumors score (26), which represent parameters on PSMA PET, posttherapy SPECT/CT reflects changes through the treatment course after each cycle of radiopharmaceutical administration and does not require the administration of an additional dose of radioisotope. Additionally, we demonstrated that posttherapy SPECT/CT can provide additional prognostic information early in the course of treatment (after cycle 2), which differs from Prostate Cancer Working Group 3 criteria (12 wk). Although these criteria are validated for clinical trial use using CT and bone scans, they are rarely used in routine clinical practice. However, future studies should evaluate the value of findings on posttherapy SPECT compared with PSMA PET.

Early identification of poor prognostic markers, particularly on images taken just after the second dose of [¹⁷⁷Lu]Lu-PSMA-617 allows for treatment personalization (27,28). It may be beneficial to consider response-adapted deintensification and intensification of treatment. Pausing treatment is useful for patients who exhibit loss of PSMA target due to an exceptional response, whereas patients with new metastatic sites may benefit from combination treatments, such as those being evaluated in the ongoing trials of [¹⁷⁷Lu]Lu-PSMA-617 plus chemotherapy (29) or [¹⁷⁷Lu]Lu-PSMA-617 plus poly(ADP-ribose) polymerase inhibitors (8). Alternatively, the next line of treatment, such as targeted α-therapy (30), a switch to chemotherapy, trial options, or best standard of care could be offered to patients with marked disease progression.

Our study had several limitations. It was a retrospective analysis, with 2 of the trials using combination treatment with poly(ADP-ribose) polymerase inhibitors and immunotherapy, which introduced potential therapeutic heterogeneity. Although not assessed in detail in this study, manual contouring of the tumor volume was time- and resource-intensive. The use of automated processing tools for this purpose is an area of interest. Nevertheless, a significant strength of our study was the use of prospectively collected clinical trial data, with robust quality control and patient follow-up.

CONCLUSION

The presence of new metastases on posttherapy SPECT/CT images after the second cycle of therapy with [¹⁷⁷Lu]Lu-PSMA-617 was an independent prognostic biomarker for poor OS and may precede PSA progression. Identifying this subgroup of patients with mCRPC with poor outcomes may facilitate strategies for treatment intensification to improve outcomes or transition to the next treatment or to best supportive care.

DISCLOSURE

This research was presented in part at the Society of Nuclear Medicine and Molecular Imaging (SNMMI) 2024 Annual Meeting as an oral presentation. James Buteau receives support from a Prostate Cancer Foundation (PCF) Young Investigator Award and Ph.D. support through an Australian Government Research Training Program scholarship. Price Jackson received support from a PCF Young Investigator Award. Shahneen Sandhu is supported by a National Health and Medical Research Council (NHMRC) Investigator Grant and reports paid advisory board roles with Bristol Myers Squibb, Merck Sharp and Dohme, AstraZeneca, Novartis, and Merck Serono (funds go to a research fund at Peter MacCallum Cancer Centre) and institutional research grants for investigator-initiated trials from Novartis, Genentech, Amgen, AstraZeneca, Merck Serono, Merck Sharp and Dohme, Pfizer, Senwha, and Genetech. Michael Hofman acknowledged philanthropic or government grant support from PCF, CANICA Oslo Norway, Peter MacCallum Foundation, an NHMRC Investigator Grant, Movember, the Medical Research Future Fund (MRFF), the U.S. Department of Defense, and the Prostate Cancer Foundation of Australia (PCFA) and acknowledges research grant support (funds go to Peter MacCallum Cancer Centre) from Novartis, ANSTO, Bayer, Isotopia, and MIM and consulting fees for lectures or advisory boards from Janssen, MSD, AstraZeneca, Astellas, and Mundipharma. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. No other potential conflict of interest relevant to this article was reported.

KEY POINTS

QUESTION: Do qualitative and quantitative analyses of posttherapy SPECT/CT images have prognostic implications in patients with mCRPC undergoing [¹⁷⁷Lu]Lu-PSMA-617 therapy?

PERTINENT FINDINGS: The appearance of new lesions on posttherapy SPECT/CT images after cycle 2 is an independent prognostic factor for OS.

IMPLICATIONS FOR PATIENT CARE: The presence of new metastases on posttherapy SPECT/CT after cycle 2 is an independent prognostic biomarker for OS and could guide future prospective research to improve treatment strategies for patients with poor prognoses.