Abstract
Tumor mutational burden (TMB) is a biomarker that predicts response to immune checkpoint inhibitor therapy. We currently lack a comprehensive understanding of how genomic and clinical factors correlate with TMB. We used a clinicogenomic database to assess independent predictors of TMB levels. The study included 2740 prostate cancer specimens from prostate gland (51.6%), lymph nodes (14.6%), and bone (10.4%). Androgen deprivation therapy use beyond 24 mo was weakly associated with high TMB (fold-change estimate [FCE] 1.14, 95% confidence interval [CI] 1.03–1.26; p = 0.009). In comparison to the prostate gland, metastases in the bladder (FCE 1.20, 95% CI 1.02–1.42; p = 0.029), liver (FCE 1.26, 95% CI 1.10–1.43; p < 0.001), and other locations (FCE 1.26, 95% CI 1.11–1.43; p < 0.001) were associated with high TMB. Microsatellite instability high (FCE 8.46, 95% CI 6.42–11.15; p < 0.001) and intermediate (FCE 1.77, 95% CI 1.46–2.14; p < 0.001) status were associated with greater TMB. Altered genes associated with greater TMB included MLH1 (FCE 1.81; p = 0.004), MSH2 (FCE 1.87; p < 0.001), MSH6 (FCE 1.92; p < 0.001), BRCA2 (FCE 1.69; p < 0.001), CDK12 (FCE 1.40; p < 0.001), MRE11 (FCE 2.28; p = 0.016), and PALB2 (FCE 2.08; p < 0.001). Our study demonstrates that TMB is relatively stable over lines of therapies and can be used to guide treatment at diagnosis or in later lines for patients with metastatic prostate cancer.
Patient summary
The number of genetic mutations in a tumor (tumor mutational burden, TMB) may help in predicting a patient’s response to immunotherapy in advanced prostate cancer. We evaluated clinical and genetic factors that may affect TMB. We found that metastases in the bladder and liver are more likely to have high TMB than the primary tumor. Some individual genes are associated with high TMB. No prior treatment type was strongly associated with TMB, suggesting that TMB can be used to guide treatment at any time point.
These data were presented at the American Society of Clinical Oncology 2023 Genitourinary Cancers Symposium.
Keywords: Prostatic neoplasms, Biomarkers, Microsatellite instability, Genomics
Currently, immune checkpoint inhibitors (ICIs) rarely benefit patients with advanced prostate cancer, with any benefit mainly in the rare group with prostate cancer with high tumor mutational burden (TMB) [1], [2]. Factors such as microsatellite instability (MSI) status [2], [3], metastatic site [1], and prior treatment history [4] may contribute to the TMB, but their relative contributions have not been assessed. We elucidated how these factors contribute to TMB status.
We included patients with metastatic prostate cancer in the US-wide Flatiron Health-Foundation Medicine (FH-FMI) deidentified clinicogenomic database (CGDB) between January 2011 and March 2022. Retrospective clinical data were obtained from electronic health records, comprising patient-level structured and unstructured data curated via technology-enabled abstraction, and were linked to genomic data from FMI comprehensive genomic profiling tests in the FH-FMI CGDB via deidentified, deterministic matching. Deidentified clinical data, including demographics, clinical and laboratory parameters, timing of treatment exposure, and survival, originated from approximately 280 US cancer clinics (∼800 sites of care) [5]. The institutional review board approved the study protocol, which included a waiver of informed consent.
Hybrid capture-based next-generation sequencing (NGS) assays were performed on tumor specimens in Clinical Laboratory Improvement Amendments–certified, College of American Pathologists–accredited laboratories. Samples were evaluated for genomic alterations [6]. TMB was determined on up to 1.1 Mb of sequenced DNA [7]. MSI status was determined via an NGS-based and fraction-based MSI algorithm that calculates the microsatellite fraction that is unstable by analyzing >2000 loci [8]. With this method, tumors are classified as microsatellite stable (MSS), MSI-equivocal/intermediate (MSI-I), or MSI-high (MSI-H).
A prospectively declared statistical analysis plan was followed. The inclusion and exclusion criteria, potential biases, primary outcome measures, exploratory outcome measures, handling of missing data, and all methods described below were specified before analysis unless otherwise noted.
We used χ2 tests and Wilcoxon rank-sum tests to assess differences between groups. A multivariable linear model was used to assess independent prediction of TMB levels. There was no adjustment for multiple comparisons. Specimens with missing clinical values were excluded.
The final analysis cohort consisted of 2740 tissue specimens obtained from men with metastatic prostate cancer. The majority of the specimens were from the prostate. Most of the men were White and most had not received any systemic treatment (Table 1).
Table 1.
Characteristics of the patient cohort overall and by TMB group
| TMB <10 (n = 2609) | TMB ≥10 (n = 131) | Overall (n = 2740) | p value | |
|---|---|---|---|---|
| Prior ADT use, n (%) | <0.001 | |||
| Unexposed | 1545 (59.2) | 55 (42.0) | 1600 (58.4) | |
| 0–12 mo | 246 (9.4) | 13 (9.9) | 259 (9.5) | |
| 12–24 mo | 154 (5.9) | 13 (9.9) | 167 (6.1) | |
| >24 mo | 664 (25.5) | 50 (38.2) | 714 (26.1) | |
| Prior NHT use, n (%) | <0.001 | |||
| Unexposed | 2111 (80.9) | 87 (66.4) | 2198 (80.2) | |
| 0–12 mo | 280 (10.7) | 29 (22.1) | 309 (11.3) | |
| >12 mo | 218 (8.4) | 15 (11.5) | 233 (8.5) | |
| Prior Ra-223 use, n (%) | 0.364 | |||
| Unexposed | 2561 (98.2) | 130 (99.2) | 2691 (98.2) | |
| Exposed | 48 (1.8) | 1 (0.8) | 49 (1.8) | |
| Prior taxanes, n (%) | <0.001 | |||
| Unexposed | 2328 (89.2) | 100 (76.3) | 2428 (88.6) | |
| 0–6 cycles | 162 (6.2) | 18 (13.7) | 180 (6.6) | |
| >6 cycles | 119 (4.6) | 13 (9.9) | 132 (4.8) | |
| Prior ICI, n (%) | 0.001 | |||
| Unexposed | 2600 (99.7) | 128 (97.7) | 2728 (99.6) | |
| Exposed | 9 (0.3) | 3 (2.3) | 12 (0.4) | |
| Race, n (%) | 0.567 | |||
| White | 1604 (61.5) | 74 (56.5) | 1678 (61.2) | |
| Black/African American | 195 (7.5) | 9 (6.9) | 204 (7.4) | |
| Other race | 346 (13.3) | 19 (14.5) | 365 (13.3) | |
| Unknown | 464 (17.8) | 29 (22.1) | 493 (18.0) | |
| Specimen site, n (%) | 0.008 | |||
| Prostate | 1368 (52.4) | 47 (35.9) | 1415 (51.6) | |
| Bladder | 96 (3.7) | 8 (6.1) | 104 (3.8) | |
| Bone | 270 (10.3) | 14 (10.7) | 284 (10.4) | |
| Liver | 222 (8.5) | 15 (11.5) | 237 (8.6) | |
| Lung | 76 (2.9) | 3 (2.3) | 79 (2.9) | |
| Lymph node | 372 (14.3) | 28 (21.4) | 400 (14.6) | |
| Other | 205 (7.9) | 16 (12.2) | 221 (8.1) | |
| MSI, n (%) | <0.001 | |||
| MSS | 2543 (97.5) | 33 (25.2) | 2576 (94.0) | |
| MSI-H | 3 (0.1) | 80 (61.1) | 83 (3.0) | |
| MSI-I | 63 (2.4) | 18 (13.7) | 81 (3.0) | |
| Gleason score, n (%) | 0.02 | |||
| Gleason ≤6 | 87 (3.3) | 6 (4.6) | 93 (3.4) | |
| Gleason 7 | 422 (16.2) | 18 (13.7) | 440 (16.1) | |
| Gleason 8 | 462 (17.7) | 18 (13.7) | 480 (17.5) | |
| Gleason 9 | 987 (37.8) | 39 (29.8) | 1026 (37.4) | |
| Gleason 10 | 238 (9.1) | 16 (12.2) | 254 (9.3) | |
| Unknown | 413 (15.8) | 34 (26.0) | 447 (16.3) |
TMB = tumor mutational burden; ADT = androgen deprivation; NHT = novel hormone therapy; ICI = immune checkpoint inhibitor; MSI = microsatellite instability; MSS = microsatellite stable; MSI-H = microsatellite instability-high; MSI-I = microsatellite instability-intermediate.
Androgen deprivation therapy (ADT) use for >24 mo before biopsy was independently associated with higher TMB (Fig. 1). Prebiopsy treatment with novel hormonal therapy, taxanes, or radium was not associated with TMB. Bladder and liver metastasis sites were associated with modestly higher TMB. Lymph nodes did not show greater TMB than the prostate gland. According to the magnitude of the effect, MSI-H showed the strongest association with greater TMB. Individual genes associated with higher TMB included MLH1, MSH2, MSH6, BRCA2, CDK12, MRE11, and PALB2 (Fig. 1).
Fig. 1.
Clinical and genomic factors associated with tumor mutational burden (TMB). Independent associations between clinical and genomic factors and TMB were estimated using a multivariable linear model. Model estimates indicate the association with fold-change in TMB, with the dashed line indicating 1 for no effect (ie, the variable is predicted to change TMB by a factor of 1). Variables to the right are associated with higher TMB, and variables to the left with lower TMB. Genes for which the alteration had a known or likely truncating or inactivating mutation or homozygous deletion are listed. CI = confidence interval; ADT = androgen deprivation therapy; NHT = novel hormone therapy; ICPI = immune checkpoint inhibitor; MSI = microsatellite instability; MSS = microsatellite stable; MSI-H = microsatellite instability-high; MSI-I = microsatellite instability-intermediate.
As anticipated, TMB was strongly associated with MSI status and mismatch repair (MMR) genes. However, after adjustment for underlying genomics, TMB was not strongly associated with prior systemic treatment apart from long-term ADT. Certain biopsy specimen sites, including bladder and liver, had higher TMB, but overall the association was relatively weak in comparison to the association for MMR genes and MSI status.
This is the first study to characterize the relative contributions of clinical and genomic variables to TMB. Overall, a minority of patients (∼5%) with prostate cancer have high TMB [1], [2], [3]. However, it is this rare subgroup that is most likely to benefit from current-generation ICIs [2], [3]. Insight into how clinical variables, prior treatments, and genomic signatures contribute to TMB is important for a better understanding of this patient population.
Certain genomic signatures change with prior lines of therapy in advanced prostate cancer. It is thought that this is due to selective pressures as the disease state advances. Markers such as AR and MYC amplifications are enriched after lines of therapy, but aberrations in homologous recombination repair (HRR) genes, including BRCA2, are largely stable over lines of treatment [4]. In this study, after accounting for additional clinical factors and genomic associations, we found that TMB was relatively stable with prior systemic treatments apart from long-term ADT. As the magnitude of this association is relatively weak in comparison to those with genomic factors, it is possible that it reflects an increase in selective pressure due to the long time on treatment, which then leads to an increase in TMB. Nonetheless, given the lack of its strong association with any treatment, we expect that TMB can be used to guide treatment decisions whether it is measured at diagnosis or in a later-line setting.
Accounting for treatment status, select specimen sites had weakly positive associations with TMB, including bladder and liver. Notably, common sites of biopsy, including bone and lymph nodes, were not associated with TMB status. Hence, our results are largely in accordance with prior studies demonstrating that metastatic tumor tissues may have slightly higher TMB than primary tumors [1].
Select individual gene alterations are associated with greater TMB. As expected, the strongest association we observed was with MSI-H status, since the majority of high-TMB tumors are MSI-H [1]. A recent observational study reported that high TMB was associated with better ICI effectiveness in comparison to taxanes (PSA response, time to next treatment, and overall survival), while low TMB was associated with comparable or worse ICI effectiveness in comparison to taxanes [2]. Alterations in the majority of the MMR genes were also associated with higher TMB. Of the HRR genes, BRCA2, CDK12, MRE11, and PALB2 were also associated with higher TMB. Prior studies showed that TMB is higher in BRCA2-deficient prostate cancer [9]. Alterations in HRR genes such as BRCA2, CDK12, PALB2, MRE11, and FANCA may be passenger secondary alterations due to MSI-H as opposed to defects in the HRR pathway [10]. However, at face value, the residual multivariable independent associations seen in our study are consistent with an independent contribution of BRCA2 to TMB.
Our study has some limitations: Our primary objective was to study associations between clinical and genomic characteristics and TMB. Thus, associations with prognosis, survival outcomes, and treatment response are beyond the scope of our study. In addition, this was a cross-sectional study, so we did not assess longitudinal changes in TMB in individual patients. Lastly, we did not validate our findings in an independent cohort.
We recognize that TMB calculations can vary considerably by panel size, gene content, and bioinformatics filtering, and use of the only FDA-approved TMB companion diagnostic tool is a strength of our approach. This novel study facilitates a comprehensive understanding of how clinical and genomic factors affect TMB status, which is relatively stable with lines of therapy and may thus be used to guide treatment over time.
Author contributions: Helen Y. Hougen had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hougen, Graf, Punnen, Mahal.
Acquisition of data: Graf, Quintanilha, Li.
Analysis and interpretation of data: Graf, Quintanilha, Li, Ross, Lin.
Drafting of the manuscript: Hougen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Graf, Quintanilha, Li.
Obtaining funding: Mahal.
Administrative, technical, or material support: Graf, Quintanilha, Li.
Supervision: Mahal.
Other: None.
Financial disclosures: Helen Y. Hougen certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Ryon P. Graf, Julia C.F. Quintanilha, Douglas I. Lin, Gerald Li, and Jeffrey S. Ross are employees of Foundation Medicine, a wholly owned subsidiary of Roche, and have equity interest in Roche. Roche produces atezolizumab, an immune checkpoint inhibitor. Brandon A. Mahal is funded by the Prostate Cancer Foundation, the American Society for Radiation Oncology, the Department of Defense, and the Sylvester Comprehensive Cancer Center. Helen Y. Hougen and Sanoj Punnen have nothing to disclose.
Funding/support and role of the sponsor: This study was supported by Foundation Medicine, a wholly owned subsidiary of Roche, which is a for-profit company and producer of FDA-regulated molecular diagnostics. Authors employed by Foundation Medicine were involved in the design and conduct of the study, analysis and interpretation of the data, and preparation, review, and approval of the manuscript.
Acknowledgments: We thank the patients whose data made this research possible, the clinical and laboratory staff at Foundation Medicine, and the team at Flatiron Health.
Associate Editor: Guillaume Ploussard
References
- 1.Chung J.H., Dewal N., Sokol E., et al. Prospective comprehensive genomic profiling of primary and metastatic prostate tumors. JCO Precis Oncol. 2019;3:e1800283. doi: 10.1200/PO.18.00283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Graf R.P., Fisher V., Weberpals J., et al. Comparative effectiveness of immune checkpoint inhibitors vs chemotherapy by tumor mutational burden in metastatic castration-resistant prostate cancer. JAMA Netw Open. 2022;5:e225394. doi: 10.1001/jamanetworkopen.2022.5394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Abida W., Cheng M.L., Armenia J., et al. Analysis of the prevalence of microsatellite instability in prostate cancer and response to immune checkpoint blockade. JAMA Oncol. 2019;5:471. doi: 10.1001/jamaoncol.2018.5801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zurita A.J., Graf R.P., Villacampa G., et al. Genomic biomarkers and genome-wide loss-of-heterozygosity scores in metastatic prostate cancer following progression on androgen-targeting therapies. JCO Precis Oncol. 2022;6:e2200195. doi: 10.1200/PO.22.00195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Singal G., Miller P.G., Agarwala V., et al. Association of patient characteristics and tumor genomics with clinical outcomes among patients with non–small cell lung cancer using a clinicogenomic database. JAMA. 2019;321:1391. doi: 10.1001/jama.2019.3241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Frampton G.M., Fichtenholtz A., Otto G.A., et al. Development and validation of a clinical cancer genomic profiling test based on massively parallel DNA sequencing. Nat Biotechnol. 2013;31:1023–1031. doi: 10.1038/nbt.2696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chalmers Z.R., Connelly C.F., Fabrizio D., et al. Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome Med. 2017;9:34. doi: 10.1186/s13073-017-0424-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Trabucco S.E., Gowen K., Maund S.L., et al. A novel next-generation sequencing approach to detecting microsatellite instability and pan-tumor characterization of 1000 microsatellite instability–high cases in 67,000 patient samples. J Mol Diagn. 2019;21:1053–1066. doi: 10.1016/j.jmoldx.2019.06.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kensler K.H., Baichoo S., Pathania S., Rebbeck T.R. The tumor mutational landscape of BRCA2-deficient primary and metastatic prostate cancer. NPJ Precis Oncol. 2022;6:39. doi: 10.1038/s41698-022-00284-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sokol E.S., Jin D.X., Fine A., et al. PARP inhibitor insensitivity to BRCA1/2 monoallelic mutations in microsatellite instability-high cancers. JCO Precis Oncol. 2022;6:e2100531. doi: 10.1200/PO.21.00531. [DOI] [PMC free article] [PubMed] [Google Scholar]