Prediction of Resistance to ¹⁷⁷Lu-PSMA Therapy by Assessment of Baseline Circulating Tumor DNA Biomarkers

DISCUSSION

PSA nonresponders were significantly more likely to have amplifications in cancer-associated genes in the ctDNA. Though this might be viewed as an oversimplistic finding, it was a statistically significant finding in our analyses. Amplifications are well known in cancer biology and are clearly related to genomic instability, a quintessential hallmark of cancer. Of note, the methods used in these ctDNA assays may not distinguish between amplification of a specific gene and an alteration in chromosomal number (aneuploidy). Thus, the fact that the broad finding of amplifications is more common in PSA nonresponders may reflect genomic instability at a larger scale. Additional studies would be required to confirm or deny this possibility. Optimally, we would have multiple validation cohorts to confirm the findings reported here.

Two particular amplifications were distinct between these 2 groups, CCNE1 and FGFR1. The CCNE1 gene encodes for cyclin E, which has high oncogenic potential. CCNE1 overexpression has been associated with worse overall survival, progression-free survival, and distant metastasis-free survival in multiple malignancies (7). CCNE1 upregulation has been associated with platinum-based therapy resistance as shown in ovarian, endometrial, and bladder cancers (8–11). There is a significant association between CCNE1 amplification and primary platinum-based treatment resistance in ovarian cancer (10). Given that both platinum and ¹⁷⁷Lu-PSMA are DNA damage–inducing therapies, the link between ¹⁷⁷Lu-PSMA treatment and resistance seems particularly plausible. More studies are needed to verify the validity of these findings.

FGFR1 amplification is associated with upregulation of fibroblast growth factor receptor 1, which is responsible for the activation of multiple downstream oncogenic pathways, such as those associated with MAPK, AKT, and STAT signaling (12). Overexpression of fibroblast growth factor receptor 1 correlates with resistance to letrozole and cyclin-dependent kinase inhibitors in breast cancer (13,14). The utility of FGFR1 amplification as a marker of therapeutic resistance in prostate cancer needs additional exploration.

CDK12 mutations are well known to have adverse effects on prognosis in mCRPC (15). In addition, CDK12 mutations are known to be deleterious for those prostate cancer patients treated with a variety of agents, including hormonal therapies, taxane chemotherapy, and poly(adenosine diphosphate ribose) polymerase inhibitors (16,17). Thus, perhaps it is not surprising that CDK12 mutations are linked to ¹⁷⁷Lu-PSMA resistance. Patients with CDK12 mutations do poorly after a wide variety of treatments, not just ¹⁷⁷Lu-PSMA. Several serum clinical and laboratory variables were predictive of no PSA decline in this small series. Higher alkaline phosphatase and lactate dehydrogenase levels have been adverse prognostic factors in multiple prostate cancer studies (18,19). Liver metastatic disease is a well-known adverse prognostic factor in advanced prostate cancer (2,18). Thus, despite being a small series, the fact that adverse prognostic findings previously reported were independently detected here suggests that our series has similarities to other larger datasets.

It is important to emphasize that our data are likely prognostic and not predictive with regard to ¹⁷⁷Lu-PSMA–based therapies. These data should not be used to make therapeutic choices at this time. First and most importantly, these data should be regarded as preliminary, and further validation is needed. Second with regard to therapeutic choices for ¹⁷⁷Lu-PSMA versus other alternative therapies for mCRPC is that, clearly, these data do not adequately address clinical decision-making. Additional prospective trials would be required to make conclusions regarding predictive versus prognostic attributes of the biomarkers reported here.

There were several limitations to this study, including its small sample size and the fact that it took place at a single institution in which patients were treated with various PSMA agents. Although PSA changes are a well-accepted response biomarker, the more important and clinically relevant parameters related to radiographic response, time to radiographic progression, and overall survival were not assessed. Further, the sample size precluded multivariate analyses, which are most appropriate to ascertain. Interactions between PSMA PET parameters and ctDNA were not assessed. Given the degree of genetic heterogeneity, it is difficult to draw conclusions about many of the ctDNA abnormalities noted, and much larger studies would be required to better understand the prognostic and predictive importance of the individual genetic mutations in these mCRPC patients. The lack of multiple-hypothesis testing represents a limitation, and different statistical methods could potentially lead to different conclusions.

We also note that the genetic landscape assessed by Guardant360 assays is incomplete (20). This particular assay does not assess all the prostate cancer–relevant genes and furthermore has distinct limitations with regard to the detection of genetic deletions. More depth to the ctDNA analyses, or tissue-based biopsies, may yield distinct answers. Ideally, one would ascertain the importance of these findings in the context of multiple clinical and laboratory variables and use much larger patient series.

Taken together, despite these limitations, the role of ctDNA as a biomarker for response when using ¹⁷⁷Lu-PSMA–based therapies is important to assess. Such analyses are important in the prognosis for a variety of cancers and a variety of treatments. The alterations identified here seem plausible on the basis of outcomes in other cancers and when using other therapies. Clearly, larger studies are justified to better understand the relationships between PSMA-targeted radioligand therapy and somatic genetics as assessed by ctDNA.