A Systematic Review and Meta-analysis of the Effectiveness and Toxicities of Lutetium-177–labeled Prostate-specific Membrane Antigen–targeted Radioligand Therapy in Metastatic Castration-Resistant Prostate Cancer

Abstract

Context:

Castration-resistant prostate cancer (CRPC) treatment is an evolving challenge. Prostate-specific membrane antigen (PSMA)-targeted endoradiotherapy/radioligand therapy (PRLT) with small-molecule, urea-based agents labeled with the β-particle–emitting radionuclide lutetium-177 (¹⁷⁷Lu) is a promising new approach.

Objective:

In this systematic review and meta-analysis, we evaluated the efficacy and toxicity of PRLT.

Evidence acquisition:

A systematic search was performed in PubMed/Medline (last updated February 18, 2019). A total of 250 studies were reviewed, and 24 studies with 1192 patients were included in the analysis. Proportions of patients with ≥50% serum prostate-specific antigen (PSA) decrease, any PSA decrease, and any PSA increase were extracted. Proportions of patients showing any grade toxicity and those with grade 3/4 toxicities based on Common Terminology Criteria for Adverse Events (CTCAE) grading were extracted from manuscripts. Overall survival and progression-free survival were evaluated. A meta-analysis of single proportions was carried out. Furthermore, we compared the two most common PRLT agents, ¹⁷⁷Lu-PSMA with ¹⁷⁷Lu-PSMA-I&T, for effectiveness and toxicity.

Evidence synthesis:

Among the 24 included studies, 20 included data on ¹⁷⁷Lu-PSMA-617, three included data on ¹⁷⁷Lu-PSMA-I&T, and one study had aggregated data for ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T. The estimated proportion of ¹⁷⁷Lu-PSMA-617–treated patients who showed a serum PSA decrease of ≥50% with at least an 8-wk interval between therapy and PSA measurement was 0.44 (0.39; 0.50). Therapy with ¹⁷⁷Lu-PSMA-I&T demonstrated an estimated proportion of patients with ≥50% PSA reduction to be 0.36 (0.26; 0.47). The aggregate results for men treated with more than one cycle of any kind of PRLT showed an estimated proportion of 0.46 (0.41; 0.51) for PSA response ≥50%. Regarding aggregate data from all of the PRLT agents, we found that grade 3 and 4 toxicities were uncommon, with estimated proportions from 0.01 (0.00;0.04) for nausea, fatigue, diarrhea, and elevated aspartate transaminase up to 0.08 (0.05; 0.12) for anemia. There was considerable heterogeneity among the studies in the “any-grade toxicity” groups. Meta-regression showed that more than one cycle of PRLT is associated with a greater proportion of patients with ≥50% PSA reduction. Overall survival according to pooled hazard ratios (HRs) for any PSA decline was 0.29 (0.18; 0.46), and for >50% PSA reduction was 0.67 (0.43; 1.07). Progression-free survival according to a pooled HR of >50% PSA reduction was 0.53 (0.32; 0.86).

Conclusions:

The relatively high number of PSA responders alongside the low rate of severe toxicity reflects the potentially promising role of PRLT in treating CRPC. The ultimate utility of this treatment modality will become clearer as multiple prospective studies continue to accrue. In the interim, this systematic review and meta-analysis can serve as a compendium of effectiveness and adverse events associated with PRLT for treating clinicians.

Patient summary:

Prostate-specific membrane antigen–targeted endoradiotherapy/radioligand therapy (PRLT) is associated with ≥50% reduction in prostate-specific antigen level in a large number of patients and a low rate of toxicity, reflecting its potential in treating castration-resistant prostate cancer. This systematic review and meta-analysis presents as a compendium of the effectiveness and adverse events related to PRLT for treating clinicians.

1. Introduction

Prostate cancer (PCa) is the second most common cancer in men worldwide, with an annual incidence of over 1.2 million and >350 000 deaths annually [1]. In the USA, PCa is the third most common cause of cancer death among men aged 60–79 yr and the second most common cause among men >80 yr of age [2]. A subgroup of patients show evidence of disease progression while on first-line systemic androgen deprivation therapy (ADT); this group of patients are considered to have castration-resistant prostate cancer (CRPC), a diagnosis that is associated with a high economic burden on patients and the healthcare system [3,4]. Treatment of these patients remains challenging, and new therapeutic approaches are being pursued actively [5]. Current recommendations include addition of taxane-based chemotherapeutic agents to ADT; however, this approach is associated with a significant increase in side effects [6]. Prostate-specific membrane antigen (PSMA)-targeted endoradiotherapy/radioligand therapy (PRLT) with small-molecule, urea-based agents labeled with the β-particle–emitting radionuclide lutetium-177 is a promising approach that has the potential to provide improved outcomes for patients with CRPC while also being well tolerated. There are multiple ongoing phase II and III clinical trials (ClinicalTrials.gov identifiers: NCT03511664, NCT03392428, and NCT03805594) as well as a large amount of retrospective data with PRLT [7–9].

PSMA is a transmembrane zinc metalloenzyme [10] that is highly expressed in the vast majority of PCa, including CRPC and poorly differentiated disease, and has a high internalization rate into PCa cells [11]. For these reasons, PSMA has become an attractive target for developing both diagnostic agents to image PCa and therapeutic agents to treat PCa. Additionally, ¹⁷⁷Lu has favorable features that make it a good choice as a radionuclide for creating effective PRLT agents [12]. The emitted β-particles have limited tissue penetration of <2 mm, decreasing damage to healthy adjacent tissues. Lutetium-177 also emits low-energy gamma rays that can be utilized to image the tumor in real time during treatment. Its relatively long half-life of 6.73 d results in targeted cancers being exposed to the therapeutic effect for long periods of time. Different ¹⁷⁷Lu-labeled PSMA-targeted ligands have been introduced, the two most common of which are the structurally similar agents ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T (imaging and treatment). These agents can be considered in progressive CRPC after exhaustion of the approved lines of treatment [13], such as chemotherapy and newer antiandrogen agents [14], or in case of contraindication to the approved treatments [15].

As alluded to above, many of the reports on PRLT are retrospective evaluations in which endoradiotherapeutics were administered under the auspices of the “compassionate use” doctrine in Europe [16]. In addition to these retrospective studies [8], there has recently been an emergence of prospective studies from different parts of the world including Australia [15], India [17], and Austria [18]. These retrospective and prospective studies, outside of reporting on the rate of response of CRPC patients to PRLT, have provided a large volume of information on the toxicities associated with PRLT administration. Although prior systematic reviews and meta-analyses in this field have mostly focused on response to PRLT in patients with CRPC [19–22], we reasoned that a more comprehensive overview of toxicities, and the severity of those toxicities, would serve as a useful reference for clinicians who refer patients, as well as nuclear medicine specialists who administer PRLT. Therefore, we conducted a comprehensive systematic review and meta-analysis of studies reporting Lu-labeled PSMA-targeted PRLT treatment outcomes with added focus on associated toxicities. In addition, we evaluated possible covariate contributions in the observed heterogeneity among studies, by conducting a meta-regression analysis.

2. Evidence acquisition

2.1. Literature review

This work was carried out according to the PICo method for clinical systematic reviews [23]. We searched PubMed/Medline for the following keywords: (177-Lu OR 177Lu OR Lu-177 OR Lutetium-177 OR theranostic OR theranostics) AND PSMA. A total of 250 entries were collected on February 18, 2019. All filters in PubMed were turned off, and there was no limit placed regarding the date of publication. The articles were reviewed against our inclusion and exclusion criteria: all retrospective or prospective studies of PRLT in humans, including single- or double-arm studies as well as randomized and nonrandomized trials were included in this meta-analysis. Only English-language articles were included. All studies evaluated a ¹⁷⁷Lu-labeled, small-molecule PRLT ligand in patients with CRPC. For purposes of this analysis, treatment response was defined as a binary outcome of serum prostate-specific antigen (PSA) decline of ≥50%. We also collected the proportion of patients showing any level of PSA decrease as well as any level of PSA increase as additional binary outcomes. Among the studies that utilized PRLT in multiple cycles, when serum PSA alterations during the overall study time were available, the different cycle changes were ignored and only the overall response rate was included in the analysis [8,24]. When different sets of serum PSA levels were reported for different cycles in one article (without providing the overall study results), the highest number for each category (PSA decrease ≥50%, any PSA decrease, and PSA increase) was selected for a meta-analysis [25–27].

Toxicities were extracted based on the Common Terminology Criteria for Adverse Events (CTCAE) grading system [28]. We categorized the toxicities into two groups: any-grade CTCAE toxicity and grade 3/4 CTCAE toxicities (ie, those with higher severity). Where applicable, we only included new toxicities that were attributed to the therapy and excluded baseline abnormalities. In papers where toxicities were not reported based on the CTCAE grading system, an estimate of the CTCAE grading was determined based on the description of the complication or further related information provided in the study whenever possible. If a toxicity was not provided based on the CTCAE grading system and no relevant clinical information was provided to allow us to generate an estimate of CTCAE grading, the number of patients manifesting that toxicity was included in the meta-analysis for all CTCAE grades, with the CTCAE grade 3/4 data blank for that study; for example, information regarding those who had fatigue was not provided in terms of CTCAE criteria in one manuscript [18]. Additional details regarding the included studies can be found in Table 1.

Table 1 – Summary of the studies included in the meta-analysis.

Number	First author	Year	Country	Study type	PRLT agent	Toxicity information included	Number of patients treated	Recruitment interval
1	Ahmadzadehfar [47]	2015	Germany	Retrospective	177Lu-PSMA-617	Yes	10	November 2013 through January 2014
2	Ahmadzadehfar [48]	2017	Germany	Retrospective	177Lu-PSMA-617	No	100	December 2014 through March 2017
3	Ahmadzadehfar [49]	2017	Germany	Retrospective	177Lu-PSMA-617	Yes	49	NA
4	Ahmadzadehfar [27]	2017	Germany	Retrospective	177Lu-PSMA-617	No	52	NA
5	Barber [9]	2019	Germany	Retrospective	177Lu-PSMA-I&T and 177Lu-PSMA-617	Yes	132	March 2013 through December 2016
6	Brauer [26]	2017	Germany	Retrospective	177Lu-PSMA-617	Yes	45	December 2014 through January 2017
7	Emmett [50]	2019	Australia	Prospective	177Lu-PSMA-617	Yes	14	NA
8	Fendler [51]	2017	Germany	Retrospective	177Lu-PSMA-617	Yes	15	September 2014 through May 2016
9	Ferdinandus [52]	2017	Germany	Retrospective	177Lu-PSMA-617	No	40	NA
10	Grubmuller [53]	2019	Austria	Retrospective	177Lu-PSMA-617	Yes	38	January 2015 through January 2018
11	Heck [7]	2019	Germany	Retrospective	177Lu-PSMA-I&T	Yes	100	December 2014 through August 2017
12	Hofman [15]	2018	Australia	Prospective	177Lu-PSMA-617	Yes	30	August 2015 through December 2016
13	Kesavan [54]	2018	Australia	Retrospective	177Lu-PSMA-I&T	Yes	20	November 2015 through November 2016
14	Khurshid [55]	2018	Germany	Retrospective	177Lu-PSMA-617	Yes	70	November 2014 through April 2016
15	Kletting [56]	2019	Germany	Retrospective	177Lu-PSMA-I&T	No	13	NA
16	Kratochwil [57]	2016	Germany	Retrospective	177Lu-PSMA-617	Yes	30	NA
17	Rahbar [8]	2017	Germany	Retrospective	177Lu-PSMA-617	Yes	145	February 2014 through July 2015
18	Rahbar [58]	2016	Germany	Retrospective	177Lu-PSMA-617	Yes	74	December 2014 through November 2015
19	Rahbar [25]	2018	Germany	Retrospective	177Lu-PSMA-617	No	71	December 2014 through November 2016
20	Rahbar [59]	2018	Germany	Retrospective	177Lu-PSMA-617	No	104	December 2014 through December 2016
21	Rathke [60]	2018	Germany	Retrospective	177Lu-PSMA-617	Yes	40	NA
22	Scarpa [18]	2017	Austria	Prospective	177Lu-PSMA-617	Yes	10	NA
23	Yadav [17]	2017	India	Prospective	177Lu-PSMA-617	Yes	31	2014–2016
24	Zang [46]	2019	China	Prospective	177Lu-PSMA-617	Yes	5	November 2017 through February 2018

NA = not available; PRLT = PSMA-targeted endoradiotherapy/radioligand therapy; PSMA = prostate-specific membrane antigen.

Studies that are published in the same year by the same first author, are numbered to make comparison easier. The numbering is kept consistent during the whole manuscript including figures.

2.2. Meta-analysis

A meta-analysis of single proportions was performed for the proportion (95% confidence interval [CI]) of patients demonstrating a PSA decrease of ≥50%. The same approach was used to determine the estimated proportions of patients with any PSA response, as well as those who showed any increase in PSA. We determined the proportion of patients showing any toxicities based on the CTCAE criteria related to PSMA as well as proportion of patients showing adverse effects with CTCAE grade 3 or 4. All available results were gathered for ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T, and the results were analyzed in aggregate as well as individually for ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T (in retrospective studies in which more than one agent was utilized, the data for the individual agents were separated whenever possible). One study did not differentiate between the use of ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T [9]; data from that study were utilized in the aggregated analysis only. Logit transformation with the inverse variance method was used to perform a meta-analysis of proportions. The inverse variance method was used because of the large number of studies with zero number of adverse effects, specifically for grade 3 or 4 toxicities. We performed the meta-analysis based on a random-effect model. I² was calculated to quantify the heterogeneity. Forest plots were created with “meta” package in R. Moreover, PSA response and complications between ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T were compared with SPSS version 25. Independent sample t test was performed to compare the mean of the outcome proportions of participants between ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T.

2.3. Evaluation of bias

Funnel plots were used for the evaluation of publication bias [29].

2.4. Evaluation of heterogeneity

To evaluate the contribution of possible covariates in the heterogeneity, a meta-regression analysis was performed with “meta” package version 4.9–5 for study outcomes showing significant heterogeneity in the meta-analysis, with significant p value for χ² test of heterogeneity. The following covariates were selected as possible contributing factors to the heterogeneity: prospective versus. retrospective studies, studies performed in economically developing countries versus developed countries, radioligand type (¹⁷⁷Lu-PSMA-617 vs non–¹⁷⁷Lu-PSMA-617), number of PRLT cycles (therapy with one cycle vs more than one cycle), and the time interval between PRLT and PSA evaluation (<8 vs ≥8 wk). To reduce the risk of data dredging, we performed meta-regression for variables that included more than eight studies. A p value of <0.05 was considered statistically significant. Proportions are reported, followed by 95% CIs in brackets.

In order to provide an idea of the diversity among prior therapies and also variables related to tumor burden, we provided the percentage of patients who received different types of therapies (prostatectomy, abiraterone, enzalutamide, chemotherapy including docetaxel or cabazitaxel, and ²²³Ra radiation therapy) and also the percentage of patients showing Gleason scores of ≤7, Gleason scores of >7, as well as local recurrence and metastasis status (lymph node, bone, and visceral, with detailed information regarding liver and lung wherever available) prior to PRLT in each study whenever available.

2.5. Evaluation of overall survival and progression-fee survival

The overall survival (OS) evaluation was based on the HRs of OS according to any PSA decline and a PSA decline of >50%. Progression-free survival (PFS) assessment was based on HRs according to >50% PSA decline. We extracted the estimated of HRs and 95% CI whenever provided by the authors. Regarding the studies that provided only Kaplan-Meier curves without readily available HR and 95% CI, GetData Graph Digitizer (http://www.getdata-graph-digitizer.com/) was used to read the graphical representations, and estimated HR and 95% CI were calculated based on a previous methodology [30]. A survival analysis was performed with review manager version 5.3 (The Cochrane Collaboration, Copenhagen, Norway). The rest of the meta-analysis was performed with R version 3.4.2 (2017-09-28) [31] and “meta” package version 4.9–5 [32].

3. Evidence synthesis

3.1. Results

3.1.1. Search results

A total of 250 articles were reviewed individually against the inclusion and exclusion criteria (Fig. 1). Twenty-nine articles met all the criteria. Five articles were removed from the meta-analysis because of an overlap of the analyzed patients with follow-up publications from the same institution [24,33–36]. As such, a total of 24 studies including 1192 patients were used in the meta-analysis (Table 1); of these studies, 20 provided separate information for ¹⁷⁷Lu-PSMA-617 (927 patients), three used ¹⁷⁷Lu-PSMA-I&T (133 patients), and one (132 patients) used a mix of ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T and reported aggregate data. None of the studies included a control group. Three studies were prospective [15,17,18], and the remaining studies were retrospective. The country where the study was conducted, the prospective versus retrospective approach, and toxicity information availability are included in Table 1. Patients in all studies had metastatic CRPC prior to PRLT, except for the study of Barber et al [9], where 39% of the patients were M0 and 37% were N0 prior to therapy.

Fig. 1 –

Flow chart of the systematic review. PRISMA = Preferred Reporting Items for Systematic Review and Meta-analyses; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen.

3.1.2. PSA response meta-analysis

Treatment response was determined based on ≥50% decline in serum PSA levels, in accordance to prior publications regarding PCa treatment [37–39]. Figures 2 and 3 show the forest plots for the serum PSA response metrics within this meta-analysis.

Fig. 2 –

Forest plots demonstrating the proportions of patients showing ≥50% decrease in PSA (A) after therapy with ¹⁷⁷Lu-PSMA-I&T or ¹⁷⁷Lu-PSMA-617, (B) after more than one cycle of therapy with ¹⁷⁷Lu-PSMA-I&T or ¹⁷⁷Lu-PSMA-617, (C) after therapy with ¹⁷⁷Lu-PSMA-617, and (D) after therapy with ¹⁷⁷Lu-PSMA-I&T. CI = confidence interval; df = degree of freedom; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen; RLT = radioligand therapy.

Fig. 3 –

Forest plots demonstrating the proportions of patients showing (A) any decrease in PSA after therapy with ¹⁷⁷Lu-PSMA-I&T or ¹⁷⁷Lu-PSMA-617, (B) any decrease in PSA after therapy with ¹⁷⁷Lu-PSMA-617, and (C) any increase in PSA after therapy with ¹⁷⁷Lu-PSMA-I&T or ¹⁷⁷Lu-PSMA-617. CI = confidence interval; df = degree of freedom; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen.

Among the studies that evaluated PSA ≥8 wk after the first PRLT, the estimated proportion of ¹⁷⁷Lu-PSMA-617–treated patients who showed a serum PSA decrease of ≥50% was 0.44 (0.39; 0.50; Table 2). Those treated with ¹⁷⁷Lu-PSMA-I&T showed an estimated proportion of 0.36 (0.26; 0.47) for a serum PSA decrease of ≥50%. No statistically significant difference was noted between response to ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T. The details of the meta-regression analysis are provided under “Evaluation of heterogeneity” in the Results section.

Table 2 – Serum PSA alterations after PRLT.

PRLT ligand	Serum PSA response metric	No. of studies	No. of patients	Estimated proportion	95% Confidence interval	I2 (%)	p value (χ2 test of heterogeneity)
177Lu-PSMA-617	≥50% PSA decrease (more than one cycle of PRLT)	16	657	0.47	0.41; 0.53	50	0.02
	Any PSA decrease	16	783	0.70	0.66; 0.75	43	0.04
	PSA increase	5	125	0.28	0.20; 0.36	0.0	0.82
177Lu-PSMA-I&T	≥ 50% PSA decrease	3	133	0.36	0.26; 0.47	18	0.29
177Lu-PSMA-I&T or 177Lu-PSMA-617	≥ 50% PSA decrease	23	1161	0.41	0.36; 0.47	67	0 < 0.01
	≥ 50% PSA decrease (only studies with ≥8 wk between PSA measurement and PRLT)	20	1073	0.44	0.40; 0.49	48	<0.01
	Any PSA decrease	17	796	0.71	0.66; 0.75	42	0.03
177Lu-PSMA-I&T or 177Lu-PSMA-617	PSA increase	6	138	0.27	0.20; 0.35	0.0	0.79

PRLT = PSMA-targeted endoradiotherapy/radioligand therapy; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen.

3.1.3. Toxicity meta-analysis

The proportion of patients who showed toxicities after PSMA therapy is shown in Table 3 as well as in Supplementary Figures 1–10.

Table 3 – Meta-analysis of the toxicities after PRLT.

Toxicity	PRLT agent	CTCAE grade 3 or 4 toxicity					CTCAE any-grade toxicity
		No. of studies, patients	Estimated proportion	95% CI	I2 (%)	p value (χ2 test of heterogeneity)	No. of studies, patients	Estimated proportion	95% CI	I2	p value (χ2 test of heterogeneity)
Anemia	177Lu-PSMA-I&T and 177Lu-PSMA-617	13, 718	0.08	0.05;0.12	42	0.06	11, 629	0.28	0.17; 0.42	88	<0.01
Diarrhea	177Lu-PSMA-I&T and 177Lu-PSMA-617	4, 244	0.01	0.00; 0.04	0.0	0.71	4, 244	0.06	0.02; 0.14	49	0.12
Elevated ALT	177Lu-PSMA-I&T and 177Lu-PSMA-617	4, 344	0.02	0.00; 0.10	55	0.08	4, 344	0.18	0.03; 0.62	96	<0.01
Elevated AST	177Lu-PSMA-I&T and 177Lu-PSMA-617	4, 45	0.01	0.0; 0.04	11	0.34	4, 345	0.21	0.12; 0.35	80	<0.01
Fatigue	177Lu-PSMA-I&T and 177Lu-PSMA-617	6,354	0.01	0.00; 0.04	0.0	0.68	7,364	0.25	0.16; 0.37	71	<0.01
Leukopenia	177Lu-PSMA-I&T and 177Lu-PSMA-617	13, 720	0.04	0.03; 0.07	13	0.31	10, 582	0.28	0.22; 0.35	61	<0.01
Nausea	177Lu-PSMA-I&T and 177Lu-PSMA-617	6, 423	0.01	0.00; 0.04	0.0	0.49	6, 423	0.12	0.04; 0.30	90	<0.01
Nephropathy	177Lu-PSMA-I&T and 177Lu-PSMA-617	8, 482	0.01	0.01; 0.04	0.0	0.66	8, 482	0.13	0.07; 0.24	78	<0.01
Thrombocytopenia	177Lu-PSMA-I&T and 177Lu-PSMA-617	14, 734	0.04	0.03; 0.06	0.0	0.66	12, 645	0.23	0.16; 0.31	72	<0.01
Xerostomia	177Lu-PSMA-I&T and 177Lu-PSMA-617	10, 488	0.02	0.01; 0.04	0.0	0.60	10, 488	0.22	0.12; 0.38	87	<0.01

ALT = alanine transaminase; AST = aspartate transaminase; CI = confidence interval; CTCAE = Common Terminology Criteria for Adverse Events; PRLT = PSMA-targeted endoradiotherapy/radioligandtherapy; PSMA = prostate-specific membrane antigen.

Comparison of toxicities between those who were treated with ¹⁷⁷Lu-PSMA-617 and ¹⁷⁷Lu-PSMA-I&T failed to show any significant difference. Overall, grade 3 and 4 toxicities were uncommon, with estimated proportions from 0.01 (0.00; 0.04) for nausea, fatigue, diarrhea, and elevated aspartate transaminase to 0.08 (0.05; 0.12) for anemia. Any-grade toxicities demonstrated considerable heterogeneity (Table 3).

In regard to CTCAE grade 3 or 4 toxicities, anemia was the highest reported complication among those who received ¹⁷⁷Lu-PSMA-617 (0.19 [0.06; 0.15]) and ¹⁷⁷Lu-PSMA-I&T (0.09 [0.05; 0.16]; Supplementary Table 1).

3.1.4. Survival meta-analysis

The pooled HR for the OS of any PSA decline was 0.29 (0.18–0.46) and for patients with >50% PSA decline was 0.67 (0.43–1.07). The pooled HR for the PFS of >50% PSA decline was 0.53 (0.32–0.86; Supplementary Fig. 11).

3.1.5. Comparison of ¹⁷⁷Lu-PSMA-I&T and ¹⁷⁷Lu-PSMA-617

A comparison of the proportion of patients showing ≥50% decline in serum PSA (p = 0.41, mean difference = 0.09, 95% CI [−1.3; 0.03]) or any decrease (p = 0.27, mean difference = −1.2, 95% CI [−0.3; −0.1]) in PSA between those who were treated with ¹⁷⁷Lu-PSMA-I&T and those treated with ¹⁷⁷Lu-PSMA-617 did not show any significant difference (for ≥50% decline variable: p = 0.41, mean difference = 0.09, 95% CI [−1.3; 0.03]; or for any PSA decrease: p = 0.27, mean difference = −1.2, 95% CI [−0.3; 0.1]). Furthermore, a comparison of the proportions of patients showing any-grade toxicity or grade 3 or 4 toxicities also did not show any significant difference between those who were treated with ¹⁷⁷Lu-PSMA-617 and those who received ¹⁷⁷Lu-PSMA-I&T (any-grade anemia: p = 0.98, mean difference = 0.005, 95% CI [−0.4; 0.04]; anemia with grade 3 or 4: p = 0.55, mean difference = −0.03, 95% CI [−0.13; 0.08]; any-grade diarrhea: p = 0.99, mean difference = 0.002, 95% CI [−0.48; 0.48]; any-grade fatigue: p = 0.56, mean difference = 0.09, 95% CI [−0.30; 0.48]; grade 3 or 4 fatigue: p = 0.70, mean difference = 0.003, 95% CI [−0.008; 0.01]; any-grade nausea: p = 0.48, mean difference = 0.17, 95% CI [−0.44; 0.79]; grade 3 or 4 nausea: p = 0.70, mean difference = 0.01, 95% CI [−0.08; 0.1]; any-grade thrombocytopenia: p = 0.95, mean difference = 0.007, 95% CI [−0.27; 0.29], grade 3 or 4 thrombocytopenia: p = 0.67, mean difference = −0.01, 95% CI [−0.08; 0.05]; grade 3 or 4 thrombocytopenia: p = 0.31, mean difference = −0.08, 95% CI [−0.58; 0.41]; grade 3 or 4 xerostomia: p = 0.64, mean difference = 0.01, 95% CI [−0.05; 0.07]; and any-grade xerostomia: p = 0.67, mean difference = 0.10, 95% CI [−0.40; 0.59]).

3.1.6. Evaluation of publication bias

Funnel plots (Supplementary Fig. 12) were used for the evaluation of possible publication bias. By visual inspection, there is no subjective asymmetry among the studies except for ≥50% PSA response, any-grade anemia, leukopenia, nephropathy, thrombocytopenia, and anemia with grade 3 or 4 toxicity.

3.1.7. Evaluation of heterogeneity

A meta-regression analysis of the proportion of patients with ≥50% PSA decline after radioligand therapy, regressed against the number of cycles of radioligand therapy (one cycle vs more than one cycle), demonstrated a significantly higher proportion of responded patients among those with more than one cycle of therapy (coefficient: 0.84, p < 0.001, 95% CI [0.36; 1.32]). Moreover, a meta-regression analysis of the proportion of patients with ≥50% PSA decline after radioligand therapy, regressed against the time interval between PRLT and PSA measurement (≥8 vs <8 wk interval) demonstrated a significant association of a higher proportion of responders with therapy in the ≥8-wk interval group (coefficient: 2.20, p < 0.001, 95% CI [1.30; 3.10]; Fig. 4A and 4B). The aggregate results for men treated with more than one cycle of PRLT (¹⁷⁷Lu-PSMA-617 or ¹⁷⁷Lu-PSMA-I&T) showed an estimated proportion of 0.46 (0.41; 0.51) for PSA response ≥50% (Fig. 2B).

Fig. 4 –

Bubble plots demonstrating the proportions of patients with ≥50% PSA decline after radioligand therapy, regressed against (A) the time interval between PRLT and PSA measurement (<8- vs ≥8-wk interval) and (B) the cycles of radioligand therapy (therapy with one cycle vs more than one cycle). PRLT = PSMA-targeted endoradiotherapy/radioligand therapy; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen.

The percentage of patients who received different types of therapies (including prostatectomy, abiraterone, enzalutamide, chemotherapy including docetaxel or cabazitaxel, and ²²³Ra radiation therapy) prior to PRLT and the percentage of patients showing Gleason scores of ≤7, Gleason scores of >7, local recurrence, and metastasis status (lymph node, bone, and visceral, with detailed information regarding liver and lung wherever available) prior to PRLT are provided in Supplementary Figures 13 and 14.

3.2. Discussion

In this systematic review and meta-analysis, we have provided a comprehensive evaluation of PSA status at the end of PRLT treatment and associated toxicity rates for patients with CRPC. The aggregate data showed that approximately 46% of CRPC patients being treated with more than one cycle of PRLT with either ¹⁷⁷Lu-PSMA-I&T or ¹⁷⁷Lu-PSMA-617 have PSA reductions of ≥50%, indicating that these agents are objectively effective for this patient population. The pooled estimated proportion of patients with grade 3 or 4 adverse events for all the evaluated toxicities was <10%. The relatively high number of PSA responders alongside the low rate of toxicity reflects the potentially promising role of PRLT in treating CRPC. Moreover, our study shows that a higher proportion of patients with ≥50% PSA reduction are associated with more than one cycle of PRLT in contrast to one cycle of therapy and also with a ≥8-wk interval between PRLT and PSA measurement in contrast to <8 wk.

To put the toxicities associated with PRLT in context, recent meta-analyses have evaluated the side effects with other treatment modalities for CRPC. For instance, Marchioni et al [40] showed that the rate of grade ≥3 adverse effects after abiraterone among chemotherapy-naïve patients with CRPC ranged from 4.4% to 15.5%. In another study, it was shown that early taxane-based chemotherapy with ADT might increase the risk of grade 3–5 adverse effects compared with ADT alone. According to the authors, addition of early taxane-based chemotherapy corresponded to 405 more grade 3–5 adverse events per 1000 (95% CI 243 more to 621 more) compared with those receiving ADT alone [6]. According to the CHAARTED trial, the range of grade 3/4 adverse events among patients who received the docetaxel-containing regimen was from 0.3% in thrombocytopenia to 12.1% in neutropenia [41]. Moreover, the STAMPEDE trial showed that grade 3–5 adverse events were reported among 32% of patients who received standard of care therapy, 32% of those who received standard of care therapy plus zoledronic acid, and 52% among those who received standard of care therapy plus docetaxel [42].

In a more recent meta-analysis with a focus on immunotherapy, the incidence of fatal adverse events was estimated as 3.2% [43]. Considering these numbers, the relatively low rate of severe toxicity with PRLT is quite promising. The most recent comparable systematic review about PSMA-targeted therapy in PCa is provided by Yadav et al [22]. Their meta-analysis included 17 studies with 744 patients based on their search done in August 2018. They provided estimated PSA response rates for ¹⁷⁷Lu-PSMA-I&T and ¹⁷⁷Lu-PSMA-617. The overall estimated proportion of patients showing any PSA decrease was 0.75 and that of patients with >50% decrease in PSA was 0.46. Both numbers are close to our estimate (0.41 for ≥50% decrease in PSA and 0.7 for any PSA decrease). Yadav et al [22] also provided descriptive results regarding toxicities from PSMA-targeted therapies, while in the current study we provided the forest plots and meta-analyzed results for ten complications.

The low number of severe adverse effects among patients who received PRLT agents may be the result of effective targeting of the cells that express PSMA. The expression of PSMA in non-PCa tissues would suggest candidate organs that may have off-target toxicities. These noncancerous tissues include the kidney, duodenum, and parotid and submandibular salivary glands [44]. Some of the adverse effects associated with PRLT agents based on this meta-analysis might be related to the abovementioned normal tissue sites of uptake. High uptake in the salivary glands may play a role in xerostomia. Normal liver uptake has been described as moderate [44], which may explain the relatively common elevation of transaminases. Anemia, leukopenia, and thrombocytopenia are perhaps more surprising given relatively low uptake in bone marrow; however, we believe that this effect is real as we subtracted the cases showing bone marrow suppression prior to PRLT whenever the data were provided.

There are other similar molecular structures utilized for PRLT that have been evaluated in a limited number of patients. For example, one study (ten patients) utilized ¹⁷⁷Lu-iPSMA [45] and another study (four patients) used ¹⁷⁷Lu-EB-PSMA [46]. Owing to limited data for these agents, we did not include these ligands in our meta-analysis. We also did not include α-particle–emitting agents in this meta-analysis, although, as data emerge with these compounds, a similar approach for the analysis efficacy and toxicities would be of value.

There are several limitations to this analysis. One limitation is the heterogeneity among the available studies; this includes the dose of radioligand provided, the time of PSA assessment, the interval between the injection and complication, and the diversity in prior treatments received for PCa.

Meta-regression was performed to find possible contributors to heterogeneity. At least part of the heterogeneity seen among studies for ≥50% PSA reduction can be explained by the difference in the number of cycles of the studies and the time interval between PRLT and PSA measurement. Our results showed that in contrast to studies with one cycle of PRLT, studies with more than one cycle are associated with a higher proportion of patients who responded to therapy. Moreover, there were a higher proportion of responders in studies with a ≥8-wk interval between PRLT and PSA measurement. A meta-analysis of the proportion of patients with ≥50% PSA reduction after more than one cycle of PRLT results in improved heterogeneity (I² decrease from 67% to 51%). Moreover, limiting the studies to those with at least an 8-wk interval between PRLT and PSA evaluation resulted in improved heterogeneity (I² decreased from 70% to 52%). We noticed considerable heterogeneity among studies for any-grade toxicities. The meta-regression failed to show the evaluated factors, namely, retrospective versus prospective studies, the economic situation of the countries, and the specific type of therapeutic agent (¹⁷⁷Lu-PSMA-617 vs ¹⁷⁷Lu-PSMA-I&T) as a contributing factor. However, a possible explanation can be the different time intervals of reporting the complication, although many studies did not provide the interval between noticing the toxicity and PRLT initiation. The other possible contributor to heterogeneity is prior therapies, including chemotherapies and new antiandrogen agents that were distributed diversely across studies. The tumor burden prior to therapy is another possible contributor to the heterogeneity. Variables such as Gleason score and metastatic status may affect the way patients responded to PRLT. The details of the proportion of patients with different therapies and burden scores prior to PRLT are provided in Supplementary Figures 13 and 14, and Supplementary Tables 3 and 4.

There are intrinsic limitations to the single-arm designs of the studies that underpin our findings. All studies were single arm; as a result, none of the included studies was randomized. Promisingly, among the studies that have been included in this meta-analysis, three were prospective studies [15–18]. According to ClinicalTrials.gov, there are 15 ongoing trials examining the role of ¹⁷⁷Lu-labeled PSMA-targeted agents in CRPC, among which three are double-arm studies. The control group in a multicenter study, which is recruiting 750 patients from several countries in North America and Europe, receives the best supportive/standard of care therapy, in contrast to the trial group that is being treated with ¹⁷⁷Lu-PSMA-617. The study is expected to be completed by May 2021 (ClinicalTrials.gov identifier: NCT03511664). There is another major ongoing trial from Australia that compares ¹⁷⁷Lu-PSMA-617 with cabazitaxel in a randomized study with 200 patients and is expected to be completed by October 2021 (ClinicalTrials.gov identifier: NCT03392428). Moreover, there is a nonrandomized sequential study of ¹⁷⁷Lu-PSMA-617 with pembrolizumab as an immunotherapeutic agent that is estimated to start in May 2019 (ClinicalTrials.gov identifier: NCT03805594). The results of this meta-analysis provide a comprehensive overview of the available data regarding the effectiveness and toxicities related to PRLT in patients with CRPC, and can be integral to clinician decision-making and appropriate counseling of patients until the results of the ongoing trials become available and we can further refine our under-standing of this evolving field.

4. Conclusions

The relatively high number of PSA responders alongside the low rate of severe toxicity indicates the potentially promising role of PRLT in treating CRPC. This systematic review and meta-analysis can serve as a compendium of effectiveness and adverse events for treating clinicians to balance the risks and benefits of this novel therapy for CRPC.