Article Text

Original research
Analysing the tumor transcriptome of prostate cancer to predict efficacy of Lu-PSMA therapy
  1. Analena Handke1,
  2. Claudia Kesch1,
  3. Wolfgang Peter Fendler2,
  4. Tugce Telli2,
  5. Yang Liu3,
  6. Alexander Hakansson3,
  7. Elai Davicioni3,
  8. Jason Hughes3,
  9. Hong Song4,
  10. Katharina Lueckerath2,5,
  11. Ken Herrmann2,
  12. Boris Hadaschik1 and
  13. Robert Seifert2
  1. 1Department of Urology, University of Duisburg-Essen and German Cancer Consortium (DKTK)-University Hospital Essen, Essen, Germany
  2. 2Department of Nuclear Medicine, University of Duisburg-Essen and German Cancer Consortium (DKTK)-University Hospital Essen, Essen, Germany
  3. 3Veracyte, Inc, Decipher Biosciences Inc, Vancouver, BC, Canada
  4. 4Division of Nuclear Medicine and Molecular Imaging, Department of Radiology, Stanford University School of Medicine, Stanford, California, USA
  5. 5Department of Molecular and Medical Pharmacology, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, USA
  1. Correspondence to Dr Analena Handke; analena.handke{at}


Rationale 177Lu-PSMA ([177Lu]Lutetium-PSMA-617) therapy is an effective treatment option for patients with prostate specific membrane antigen (PSMA)-positive metastatic castration-resistant prostate cancer, but still shows a non-responder rate of approximately 30%. Combination regimes of programmed death-ligand 1 (PD-L1) inhibition and concomitant 177Lu-PSMA therapy have been proposed to increase the response rate. However, the interplay of immune landscape and 177Lu-PSMA therapy efficacy is poorly understood.

Methods Between March 2018 and December 2021, a total of 168 patients were referred to 177Lu-PSMA therapy in our department and received a mean total dose of 21.9 GBq (three cycles in mean). All patients received baseline PSMA positron emission tomography to assess the PSMA uptake. The histopathological specimen of the primary prostate tumor was available with sufficient RNA passing quality control steps for genomic analysis in n=23 patients. In this subset of patients, tumor RNA transcriptomic analyses assessed 74 immune-related features in total, out of which n=24 signatures were not co-correlated and investigated further for outcome prognostication.

Results In the subset of patients who received 177Lu-PSMA therapy, PD-L1 was not significantly associated with OS (HR per SD change (95% CI) 0.74 (0.42 to 1.30); SD: 0.18; p=0.29). In contrast, PD-L2 signature was positively associated with longer OS (HR per SD change 0.46 (95% CI 0.29 to 0.74); SD: 0.24; p=0.001; median OS 17.2 vs 5.7 months in higher vs lower PD-L2 patients). In addition, PD-L2 signature correlated with PSA-response (ϱ=−0.46; p=0.04). The PD-L2 signature association with OS was significantly moderated by L-Lactatdehydrogenase (LDH) levels (Cox model interaction p=0.01).

Conclusion Higher PD-L2 signature might be associated with a better response to 177Lu-PSMA therapy and warrants further studies investigating additional immunotherapy. In contrast, PD-L1 was not associated with outcome. The protective effect of PD-L2 signature might be present only in men with lower LDH levels.

  • Prostatic Neoplasms
  • Radiotherapy
  • Tumor Biomarkers

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • [177Lu]Lutetium-PSMA-617 (Lu-PSMA) radioligand-therapy (RLT) is a novel therapeutic option targeting the prostate specific membrane antigen (PSMA). PSMA expression can be non-invasively measured in all metastases with positron emission tomography-imaging to improve the selection of patients for Lu-PSMA therapy. Patients with end-stage prostate cancer (metastatic castration-resistant prostate cancer (mCRPC)) are regularly treated with Lu-PSMA therapy, but still approximately 30% of those patients are non-responders.


  • We analyzed immune-related gene expression signatures in the primary tumor tissue to better understand why Lu-PSMA therapy is not effective in some patients with mCRPC. We could show that programmed death-ligand 2 (PD-L2) gene expression signature levels were associated with overall survival and biochemical response in patients with mCRPC treated with Lu-PSMA, whereas PD-L1 was not significantly associated with the outcome.


  • Our findings support a possible crosslink of PD-L2 signatures and sensitivity to RLT, although the underlying molecular mechanism is currently not fully elucidated. The strong association of transcriptomic features like PD-L2 with the outcome of patients with mCRPC warrants further trials investigating the combination of immunomodulation and RLT.


Several therapeutic options for men with metastatic castration-resistant prostate cancer (mCRPC) are available, making the selection of optimal treatment challenging. Most prostate cancer variants show strong upregulation of the prostate-specific membrane antigen (PSMA). Therefore, the PSMA targeting [177Lu]Lutetium-PSMA-617 (Lu-PSMA) radioligand-therapy (RLT) is a promising therapeutic option for men with end-stage prostate cancer. The beta radiation emitting radioisotope [177Lu]Lutetium is carried by the small ligand PSMA-617, which is internalized and accumulated by PSMA expressing cells.1 Because of this selectivity for PSMA, Lu-PSMA therapy has a favorable efficacy and toxicity profile1: Lu-PSMA prolonged overall survival (OS) in men with end-stage prostate cancer compared with the best standard of care and was recently approved by the Food and Drug Administration.2 In addition, the non-responder rate to Lu-PSMA therapy is approximately 30%.

It is therefore relevant to improve the identification of patients who will likely benefit from Lu-PSMA therapy and should be referred to it. For example, the PSMA expression can be non-invasively assessed by measuring PSMA uptake with PSMA targeting positron emission tomography (PET) and PSMA uptake correlates with the dose delivered by Lu-PSMA therapy.3 4 Also low PSMA uptake is associated with unfavorable response to Lu-PSMA therapy.5 Several approaches have been proposed for improved patient selection, including the analysis of PSMA uptake prior to start of therapy, the quantification of the total tumor volume of all metastases, the history of previous chemotherapies, and the use of laboratory values.5–8 A nomogram integrating several of these features has been introduced to enable improved patient selection.9 However, the management of patients with mCRPC is still challenging partly due to the complex relationship of the tumor phenotype and the immunogenic response to it.10

It has been hypothesized that the efficacy of radiation therapy can be improved by the use of immune checkpoint inhibition through synergetic effects.11 This could be mediated by a synergistic stimulation of antigen presenting cells by the radiation-induced release of tumor epitopes and suppression of immune escape.12 In line with this, approaches combining Lu-PSMA therapy and programmed death-ligand 1 (PD-L1) checkpoint inhibition have been proposed to target mCRPC, with initial trials to enhance the efficacy of Lu-PSMA therapy currently underway (NCT03805594, NCT03658447).13,14 Therefore, there is an urgent need to better understand the immune tumor microenvironment of patients with mCRPC who are referred to Lu-PSMA therapy. Furthermore, the analysis of the immune tumor microenvironment could help in identifying patients with the best chance of responding to Lu-PSMA therapy. Biochemical response is widely used when the efficacy of Lu-PSMA therapy is evaluated and often defined as PSA decline greater than 50% from baseline; this definition is also followed in the present manuscript.15

This study retrospectively assessed the transcriptome of initial diagnostic tumor tissue biopsy of patients with mCRPC who received Lu-PSMA therapy to analyze the immune microenvironment, potentially identifying molecular features that might be negatively correlated with Lu-PSMA efficacy and could therefore be addressed by concomitant immunotherapy.

Materials and methods

Patient inclusion

We screened 168 patients who underwent a Lu-PSMA therapy at the Department of Nuclear Medicine of the University Hospital Essen, Germany, from March 2018 until December 2021 and had available clinical data (online supplemental figure 1). Of these patients, primary pathology was requested from the primary pathology institutes. A total of 34 samples were retrieved; the remaining samples were largely missing due to exceeding the legal retention period between diagnosis and start of PSMA therapy (ie, more than 10 years).

Supplemental material

Lu-PSMA therapy

The decision for Lu-PSMA therapy was made in interdisciplinary tumor boards. Patients must have had adequate organ function and hematological reserve.16 All patients received prior chemotherapy, androgen deprivation therapy, and at least one novel hormonal agent (NHA) (table 1), and all patients underwent 68Ga-PSMA-11 or 18F-PSMA-1007 PET/CT before Lu-PSMA therapy. Baseline PSMA-PET (field of view: vertex to mid thighs, ie, whole body) were used to semi quantitatively assess the level of PSMA uptake and to rule out the presence of metastases without PSMA uptake. For eligibility, PSMA uptake of all metastases must exceed the physiological PSMA uptake of the liver (or spleen in case of liver-dominant excretion of PSMA-1007). A Time-of-Flight (TOF-)based algorithm (3 iterations, 21 subsets) was used for PET image reconstruction. The PSMA-PET was repeated every two cycles of Lu-PSMA therapy to monitor treatment response. For this study, the PET images were re-analyzed. To this end, an established semi-automatic workflow for PSMA tumor volume quantification using a 50% isocontour was chosen to segment the tumor lesions.8 17 The workflow assists in the selection of all metastases in the whole-body PET acquisition. The mean tumor uptake was calculated by averaging the maximum standardized uptake value (SUVmax) of all metastases. Minimal PSMA uptake was defined as SUVmax of the metastasis with the lowest SUVmax among all metastases of a given patient. This parameter was obtained, because previous studies could show that patients with low PSMA uptake of some metastases despite high average PSMA uptake of all metastases have unfavorable outcomes, which could be biologically explained by higher inter-metastatic tumor heterogeneity.5

Table 1

Patient characteristics

If patients were eligible, Lu-PSMA therapy was performed as described previously. Briefly, 6–7.4 GBq Lu-PSMA were administered as slow intravenous administration every 6–8 weeks until adverse reaction or tumor progression up to a total number of six cycles.16 A post therapeutic whole-body scintigraphy was performed 24 hours after administration of Lu-PSMA therapy to ensure adequate distribution, rule out tumor progression and check for renal retention.

Microarray analysis for transcriptomic profiling

Tissue samples were requested from the primary institute of pathology for each patient who gave consent to re-analysis of specimens. For biopsy tissue, one formalin-fixed paraffin embedded block of the diagnostic needle-core biopsy with at least 0.5 mm in total linear length of tumor tissue and with the highest-grade group was selected. If the biopsy sample was not available, the radical prostatectomy block with the highest-grade group and tumor volume (≥0.5 mm2) was selected. Tumor tissue was macrodissected from non-neoplastic areas using a representative H&E slide as a guide after a pathologist had marked the region of interest.

RNA was extracted from formalin-fixed paraffin embedded tumor tissue and underwent cDNA amplification, oligonucleotide microarray hybridization and microarray quality control in a Clinical Laboratory Improvement Amendments-certified laboratory (Veracyte, San Diego, California, USA) as previously described.18 At least 50 ng of extracted RNA (Qiagen RNeasy FFPE kit) was used for cDNA amplification (Tecan Ovation FFPE kit) and 3.5–7.0 µg of cDNA were used for hybridization to Affymetrix Human Exon 1.0 ST microarrays (Thermo Fisher, Santa Clara, California, USA). Genome-wide expression profile data were first preprocessed and normalized using the Single Channel Array Normalization algorithm.19 Precomputed and locked gene expression signatures (n=459) were retrieved from the Genomics Resource for Intelligent Discovery (GRID) database (accessed on October 19, 2022).20 Given the small sample size, analyses were focused on a subset of n=74 tumor immune microenvironment-related gene expression signatures (online supplemental table 1) to examine the hypothesis that the interplay between tumor and immune cells is associated with the efficacy of radionucleotide therapy.11 In order to further reduce the chances of false discovery a Spearman correlation filter (<0.70) was applied to trim highly correlated (ie, biologically redundant) immune microenvironment signatures (online supplemental table 1) for analysis leaving a set of 24 for further analysis (figure 1A).

Figure 1

Association of immune marker signatures with PSA response and OS. (A) The correlations between the 24 non-redundant (ie, those without high Spearman correlations among each other) set of evaluated tumor immune microenvironment gene expression signatures (ρ is shown), (B) OS of the group with tumor immune gene expression signature analysis is not significantly different from the overall cohort of Lu-PSMA therapy treated patients at our institution and (C) the cDNA yield from RNA extracted from tumor samples, best PSA response and relative expression signature level (z-scores, higher values indicate higher signature) of immune-related gene signatures are shown correspondingly to identify patterns of signatures with regard to PSA response. Patients with higher PD-L2 signature (above the median) have longer OS time (D) and greater PSA responses (E). Lu-PSMA, [177Lu]Lutetium-PSMA-617; mo., months; OS, overall survival; PD-L1, programmed death-ligand 1; PSA, prostate-specific antigen.

Figure 2

Univariable Cox regression for OS prognostication. Medians, ranges, and HRs per SD unit change estimated from univariable Cox regression for OS for 24 tumor immune microenvironment gene expression signatures (after trimming highly co-linear signatures) after Lu-PSMA therapy. HLA-C, human leukocyte antigen c; IDO, Indoleamine 2, 3-dioxygenase; IL-8, Interleukin 8; Lu-PSMA, [177Lu]Lutetium-PSMA-617; NK, Natural killer cells; OS, overall survival; PD1, programmed cell death protein; PD-L1, programmed death-ligand 1, TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM3, T-cell immunoglobulin mucin-3.

Statistical analysis

Univariable and multivariable Cox proportional hazards models were used to assess the relationship between marker values and OS, with a Bonferroni correction applied to minimize the risk of alpha error accumulation (adjusted alpha level: 0.002, (0.05/24)). Continuous signatures were normalized to SD units to facilitate comparison of effect sizes across markers with varying ranges. The Kaplan-Meier method was used to visualize survival differences between patients with signature scores dichotomized by their median value. For parameters identified by the Cox analysis, log-rank tests were used to determine the statistical significance of differences in survival curves. The maxstat package of R was used to find the optimal cut-off for visualizing the OS with regard to PET parameters.21 A heatmap of model estimated 12-month risk of death illustrates the relationship between PD-L2 signature and L-Lactatdehydronease (LDH). Unpaired Wilcoxon test was used to compare blood values between responders and non-responders.


Patient characteristics

The tissue samples from 34 patients with mCRPC (31 biopsy and 3 RP specimens) were retrieved (online supplemental figure 1 Consolidated Standards of Reporting Trials diagram). Among these, seven samples failed pathological review (insufficient tumor content remaining in the block), one sample failed cDNA amplification (RNA too degraded to amplify) and three failed microarray quality control (gene expression profiles with low signal to noise). This left 23 patient transcriptomes for analysis. Detailed patient characteristics for the total cohort are shown in table 1. The metastatic extent of patients who underwent transcriptomic analysis (table 2) and separated details for responder (ie, PSA decline >50 %) and non-responder (ie, PSA decline <50 %) (table 3) are also shown. The proportion of PSA responders and non-responders was balanced. The median time from prostate biopsy until start of Lu-PSMA therapy was 50 months (range: 14–126). The OS of all men treated with Lu-PSMA therapy between March 2018 and December 2021 in Essen was 10.2 months (patients with follow-up: n=160); OS was not statistically significant between cohorts with (OS=7.3 months) or without transcriptomic analysis (OS=11.2 months) (p=0.25) (figure 1B).

Table 2

Metastatic extent of patients who underwent the transcriptomic analysis

Table 3

Characteristics of patients who underwent the transcriptomic analysis

We were able to retrospectively collect and successfully process expression analysis samples from 23 patients, who are shown here. Data is presented separately for responding (defined as at least 50% PSA response to Lu-PSMA therapy) and non-responding (defined as maximum 50% PSA response to Lu-PSMA therapy) patients. The Gleason score of one patient of the non-responder group (ie, PSA decline <50%) was missing. In addition, 2 of the 23 patients had only survival follow-up data available, but no PSA follow-up. Both patients received only one cycle of RLT because of symptomatic progression and died 1 and 6 months after the end of therapy, respectively. These 2 patients were included in the OS analysis of the cohort (n=23), but not stratified according to PSA response in this table.

Association of transcriptomic features with overall survival

Higher PD-L2 and T-cell immunoglobulin mucin-3 (TIM3) signatures were significantly associated with better OS (PD-L2 HR per SD change 0.46 (95% CI 0.29 to 0.74); p=0.001; SD: 0.24) and TIM3 HR per SD change 0.56 (95% CI 0.33 to 0.95); p=0.03; SD: 0.15), while higher CD27 (HR per SD change 1.93 (95% CI 1.06 to 3.48); p=0.03; SD: 0.23) signature was significantly associated with worse OS (figures 1 and 2). After applying the Bonferroni correction, only the PD-L2 signature in tumor tissue remained significant. Therefore, further analysis was performed with PD-L2 alone.

Association of PD-2 levels with the response and outcome

PD-L2 was significantly correlated with PSA decline in response to Lu-PSMA therapy (ϱ: −0.46; p=0.037). Patients with higher PD-L2 signature (stratified by median) had longer OS than those with lower signature (median OS: 17.2 vs 5.7 months; log-rank p=0.008) (figure 1). For comparison, there was no significant association of PD-L1 with PSA response (ϱ:0.11; p>0.05) or OS (HR 0.74 (95% CI 0.42 to 1.30]; p=0.29) measurable.

PSMA PET derived parameters

To further investigate the interplay of PD-L2 signature, PSMA uptake and total tumor volume, all PSMA-positive tumor manifestations were delineated on baseline and interim (after the first two cycles of Lu-PSMA-RLT) PSMA PET/CT. Average PSMA uptake of all metastases and minimal PSMA uptake as well as the total tumor volume were measured. Neither the log transformed average PSMA uptake (mean SUVmax of all metastases) at baseline nor the total tumor volume was a significant prognosticator of OS time in univariable regression (HR 0.31 (95% CI 0.07 to 1.38); p=0.12 and HR 1.16 (95% CI 0.90 to 1.68); p=0.42, respectively). However, the log transformed minimal PSMA uptake was a significant prognosticator of OS (HR 0.01 (95% CI 0.00 to 0.87); p<0.05). The baseline tumor volume was significantly associated with PSA response (ϱ:−0.44; p<0.05), whereas neither the average PSMA (FOLH1) gene expression nor the average PET-derived PSMA uptake was significantly correlated with PSA response (ϱ:0.02; p=0.91 and ϱ:−0.31; p=0.17). PD-L2 signature was positively correlated with baseline average PSMA uptake (ϱ:0.45; p=0.04) but not with baseline minimal PSMA uptake (ϱ:0.15; p=0.5, figure 3B).

Figure 3

Association of OS and PSMA PET derived parameters with PD-L2. The response in PSMA PET derived tumor volume between baseline and interim PET (after two cycles) are shown together with OS time (in months; arrowheads indicate censored patients which are still alive) ((A); not all patients had interim PET). PET features were log transformed; all features were shown as z-score in the heatmap analysis. PD-L2 signature was positively correlated with the mean PSMA PET uptake of metastases, however, no significant association with the minimal PSMA uptake of metastases was observed (B). The OS is shown with regard to low and high PSMA-PET derived tumor volume or mean PSMA uptake, respectively. The cut-off to differentiate low and high was calculated to maximize the difference in OS (C). The heatmap of model-estimated 12-month risk of death illustrates the relationship between PD-L2 and LDH. In higher LDH patients, the risk is more homogenous across PD-L2 signature levels, while in lower LDH patients there is more prognostic value to the PD-L2 signature (D). LDH, L-Lactatdehydrogenase; mo., months; OS, overall survival; PD-L2, programmed death-ligand 2; PET, positron emission tomography; PSMA, prostate-specific membrane antigen; SUVmax, maximum standardized uptake value.

Multivariable Cox regression

We examined the association of PD-L2 signature with serum LDH. LDH (log transformed) was a significant negative predictor of OS in this cohort (HR 2.35 (95% CI 1.19 to 4.26); p=0.001). This parameter was chosen prior to other variables as a marker of metastatic disease burden given its previously reported role as a survival biomarker for Lu-PSMA therapy and the non-significant association of PSMA PET derived tumor volume and OS in this cohort.22 In multivariable Cox regression analysis of baseline LDH (log transformed) and PD-L2×LDH interaction, PD-L2 signature was a positive OS predictor (p=0.005), whereas PD-L2×LDH interaction was associated with shorter OS (p=0.011) (see figure 3D for details).


The interplay of the tumor immune microenvironment and the response to Lu-PSMA therapy is poorly elucidated but could be relevant to understand treatment failure or early disease progression of patients treated with Lu-PSMA therapy. Therefore, as an initial approach the present study analyzed the tumor immune microenvironment gene expression signatures of patients with mCRPC who received Lu-PSMA therapy. Of note, the gene expression signature analysis was conducted in the primary tumor transcriptomes from diagnostic or surgical resection specimens. After controlling for multiple comparisons, we found PD-L2 gene expression signature levels determined in the primary tumor tissue were associated with OS in patients with mCRPC treated with Lu-PSMA. PD-L1 levels, however, were not associated with the outcome.

Naively, the expression of PD-L1 and PD-L2 as markers of immune escape should be negatively correlated with patient outcome; however, the role of PD-L2 is not fully understood to date.23 The present analysis found that higher signatures of PD-L2 (ie, low expression of PD-L2) was associated with longer OS with radionucleotide therapy. In line with our finding, preliminary evidence shows that PD-L2 is a poor prognostic factor in prostate cancer in general but suggests an association of higher PD-L2 signature with benefit in case of radiation therapy.24 The findings of this study support a possible favorable effect of PD-L2 signature broadly in the context of sensitivity to different forms of radiotherapy, although the underlying biology is currently not fully elucidated.25 Given that all patients in our cohort received RLT, the efficacy of Lu-PSMA therapy may be associated with baseline PD-L2 signature. Without the benefit of a control group, we cannot verify whether PD-L2 signature is simply a favorable prognostic factor or indeed predictive of improved response to Lu-PSMA therapy in the mCRPC setting. This emphasizes the need for randomized trials in further studies.

Previous studies indicated low levels of PD-L1 expression in prostate cancer.24 26 In line with this finding, the expression signature of PD-L1 was not associated with the outcome of patients receiving Lu-PSMA therapy in the present study. Given the currently ongoing trials investigating the benefit of accompanying administration of PD-L1 targeting checkpoint inhibition for Lu-PSMA therapy (NCT03805594, NCT03658447), the present results indicate that targeting PD-L1 might be of subordinate relevance to increase the efficiently of RLT in mCRPC.

The level of PSMA uptake is correlated with the radiation dose delivered to the cancer cells and should therefore be associated with the efficacy of PSMA-therapy.3 Corroborating this idea, the PSMA PET-derived PSMA uptake was associated with PSA response. Also, the minimal PSMA uptake of any metastasis was positively associated with the OS. Both findings are in line with previous trials, indicating that higher PSMA uptake in all metastases is associated with favorable outcomes.5 27 The biological basis of this is not fully elucidated; it was hypothesized that average PSMA uptake might indicate poorly differentiated metastases with lost ability to express PSMA and low minimal PSMA uptake despite high average PSMA uptake could indicate unfavorable inter-metastatic tumor heterogeneity.5 However, the underlying tumor biology of this association is still poorly elucidated.5 Inter alia, de-differentiation of prostate cancer cells might lead to loss of PSMA expression, but also to more aggressive tumor phenotypes.28 29

In the multivariable analysis, we observed a significant interaction between PD-L2 signature and LDH levels. This suggests that when LDH levels are high (ie, greater disease burden) higher PD-L2 signature may be associated with worse outcome. Thus, PD-L2 seems to prognosticate the outcome of patients treated with Lu-PSMA therapy in mCRPC, but only for those with lower disease burden, when LDH under Lu-PSMA therapy is low. This may indicate that LDH levels are associated with a more aggressive tumor phenotype or that the immune response against cancer loses efficacy with high tumor burden.

A limitation of this retrospective study is the small sample size, which implies a limited validity of our results. Of 168 patients treated, only 34 had tissue accessible for transcriptomic profiling, which is due to the maximum retention period for histopathological specimens of 10 years. This could also result in a limited representativeness of the cohort, as patients with less aggressive tumors who have a longer disease course (ie, more than 10 years) could not be considered. In this context, a missing control group is also a limitation that limits the significance of the results. Another potential limitation is the long interval between tumor biopsy and the multiple lines of therapy these patients were subject to and the initiation of Lu-PSMA therapy start and therefore possible changes in microenvironment that are expected. During this long period, patients have also experienced multiple therapies, which presumably have led to immunomodulatory changes in tumor tissue. Therefore, the results regarding PD-L1 and PD-L2 levels and the associated outcome of RLT should be interpreted with caution, further analyses on, for example, metastatic tissue with a temporally shorter sampling to the start of RLT are needed. However, the strong association of transcriptomic features with the outcome still suggests that immune-related gene expression signatures in the primary tumor may be useful biomarkers for further risk stratification of patients with mCRPC. Specifically, PD-L2 signature at time of diagnosis was strongly associated with improved response to Lu-PSMA therapy and survival outcomes after Lu-PSMA therapy. To our knowledge, there is no data available comparing the transcriptomic profile of primary and metastatic tissue in prostate cancer. Hence, further studies are needed to analyze the profile of metastatic tissue, also, to identify whether PD-L2 signature remains a comparable prognostic factor.


Our findings suggest crosstalk between the tumor immune microenvironment and the outcome of patients treated with Lu-PSMA therapy. Among others, higher PD-L2 signature might be associated with an improved outcome of patients treated with Lu-PSMA therapy by a currently unknown interplay of immune environment and radiation efficiency. Future studies investigating the tumor immune environment of patients receiving Lu-PSMA therapy are warranted.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statements

Patient consent for publication

Ethics approval

The study was approved by the local ethics review board of the University of Duisburg-Essen (reference number: 21-9882-BO). Participants gave informed consent to participate in the study before taking part.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors AHan: Conzeptualization, Validation, Formal analysis, Investigation, Data curation, Writing—Original draft, Writing—Review and editing, Visualization. RS: Guarantor, Conzeptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Writing—Original draft, Writing—Review and editing, Visualization. CK: Conzeptualization, Investigation, Writing—Review and editing. WPF: Data curation, Writing—Review and editing. TT: Data curation, Writing—Review and Editing. YL: Formal Analysis, Writing—Review and editing, Visualization. AHak: Formal Analysis, Writing—Review and editing, Visualization. ED: Methodology, Validation, Formal analysis, Writing—Original draft, Writing—Review and editing, Supervision, Project administration. JH: Methodology, Validation, Formal analysis, Writing—Original draft, Writing—Review and editing. HS: Validation, Formal analysis, Writing—Review and editing. KL: Data curation, Writing—Review and editing. BH: Conzeptualization, Methodology, Writing—Original draft, Writing—Review and editing, Supervision, Project administration. KH: Conzeptualization, Methodology, Writing—Original draft, Writing—Review and editing, Supervision, Project administration.

  • Funding Boehringer Ingelheim Fonds have partly provided research funding for this project.

  • Competing interests RS has received research funding from the Else Kröner-Fresenius-Stiftung outside the submitted work and research support from Boehringer Ingelheim Fonds. CK has received consulting fees from Apogepha; has received research funding from AAA/Novartis, and curie therapeutics; and has received compensation for travel from Janssen Pharmaceuticals. BH has had advisory roles for ABX, AAA/Novartis, Astellas, AstraZeneca, Bayer, Bristol Myers Squibb, Janssen R&D, Lightpoint Medical, Inc., and Pfizer; has received research funding from Astellas, Bristol Myers Squibb, AAA/Novartis, German Research Foundation, Janssen R&D, and Pfizer; and has received compensation for travel from Astellas, AstraZeneca, Bayer and Janssen R&D. ED, AHak, YL and JH are employees of Veracyte, Inc., the maker of the Decipher assay. WPF reports fees from SOFIE Bioscience (research funding), Janssen (consultant, speakers bureau), Calyx (consultant), Bayer (consultant, speakers bureau, research funding), Parexel (image review) and AAA (speakers bureau) outside of the submitted work. WPF received financial support from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, grant FE1573/3-1 / 659216), Mercator Research Center Ruhr (MERCUR, An-2019-0001), IFORES (D/107-81260, D/107-30240), Doktor Robert Pfleger-Stiftung, and Wiedenfeld-Stiftung/Stiftung Krebsforschung Duisburg. KL reports consultant fees from Sofie Biosciences and research funding from Curie Therapeutics; all outside of the current manuscript. KH reports personal fees from Bayer, personal fees and other from Sofie Biosciences, personal fees from SIRTEX, non-financial support from ABX, personal fees from Adacap, personal fees from Curium, personal fees from Endocyte, grants and personal fees from BTG, personal fees from IPSEN, personal fees from Siemens Healthineers, personal fees from GE Healthcare, personal fees from Amgen, personal fees from Novartis, personal fees from ymabs, all outside the submitted work. All authors declare no conflict of interest regarding this manuscript.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.