NEW YORK – A new targeted, cell-free DNA (cfDNA) methylation assay can detect otherwise difficult-to-diagnose castration-resistant neuroendocrine prostate cancer (CRPC-NE), with the aim of finding it early in its evolution and informing patient treatment.
Scientists at the Dana-Farber Cancer Institute and the University of Trento in Italy dubbed the assay NEMO for NEuroendocrine MOnitoring panel and plan to develop it commercially. They published their findings on Wednesday in Cancer Discovery.
Himisha Beltran, a medical oncologist at Dana-Farber and the study's senior author, explained that cancers often change their phenotype as they adapt to different treatments and that in emerging adaptation mechanism, called lineage plasticity, a tumor effectively changes identity, adopting a path toward a different cell lineage.
With respect to prostate cancer, "the most common is neuroendocrine lineage," Beltran said. "And when this happens, [the tumor] can be very aggressive, and there are limited therapies for patients. And it can be very challenging to diagnose."
As Beltran noted, CRPC-NE is an aggressive and treatment-resistant prostate cancer subtype. Treatment resistance arises through multiple mechanisms, making early detection critical to forming treatment decisions. Diagnosis can be challenging due to patient-to-patient tumor heterogeneity and the need for metastatic biopsies.
Methylation patterns, Beltran said, are the most distinguishing feature of the different prostate cancer subtypes, which led her and her colleagues to design the NEMO assay.
In their study, Beltran and her colleagues identified a set of 53 methylation sites that distinguished tumor from non-tumor cfDNA and another set of approximately 8,000 sites that distinguished the neuroendocrine and adenocarcinoma subtypes.
Based on these results, the group designed the targeted NEMO panel. NEMO quantifies tumor content and provides a "phenotype evidence score" meant to capture diverse CRPC phenotypes.
The methylation patterns differentiating tumor from non-tumor cfDNA and those differentiating different subtypes form the two "modules" of the NEMO assay. Methylation sites in the second, subtyping module contribute to the phenotype evidence score, which ranges from zero to one, where zero means total absence of evidence of CRPC-NE, and one means maximal evidence.
The team tested this scoring method in prostate cancer cell lines, in patient-derived organoids and xenografts, and in cfDNA data from patients with pathologically confirmed CRPC-Adeno and CRPC-NE. They consistently observed that confirmed cases of CPRC-NE resulted in high phenotype scores.
With the phenotype score calculated based on tissue biopsies, the group successfully identified 79 of 82 CRPC-Adeno samples and 17 of 20 CRPC-NE samples. One patient received a low phenotype evidence score despite having a metastatic tumor harboring both adenocarcinoma and neuroendocrine features. In this case, the low score was attributed to the uneven contribution of those two phenotypes. Other misclassifications were due to low tumor content, which the researchers said likely hampered classification accuracy.
The team then assessed NEMO's clinical utility by profiling plasma from participants of two completed, prospective Phase II trials, one testing the efficacy of alisertib and the other testing the efficacy of carboplatin plus docetaxel chemotherapy. Patients in these trials had clinically defined aggressive variant prostate cancer (AgAdeno) and/or pathologically-defined CRPC-NE.
"We applied the platform to two different prospective clinical trials," Beltran said, "to look at the clinical significance and performance of the assay."
From among 60 participants in the alisertib trial, they noted that patients with 10 percent or more tumor content showed poorer overall and progression-free survival than those with less than 10 percent tumor content.
Importantly, Beltran noted that all trial participants had a biopsy prior to beginning the trial, enabling her team to compare baseline plasma samples with the biopsy, as well as with clinical outcomes.
"What we found was that tumor fraction or tumor content detected by this methylation assay was prognostic, in that the phenotypic score could distinguish the biopsy-confirmed neuroendocrine patients from the other patients that had very aggressive clinical features but did not have a pathologist [who determined the tumor to be] neuroendocrine," Beltran said.
The team evaluated 41 baseline plasma samples in the carboplatin plus docetaxel trial, although not all participants in this study had prior biopsies for comparison. The presence of tumor cfDNA consistently associated with overall survival from the start of chemotherapy, and the phenotypic score successfully identified six of eight participants with CRPC-NE and identified several others whose high scores made them potentially compatible with undiagnosed CRPC-NE.
One patient in this study received a low phenotype event score despite having a metastatic tumor harboring both adenocarcinoma and neuroendocrine features, which Beltran's team attributed to the uneven contribution of the two phenotypes.
Rather than viewing this as a potential weakness of the assay, Beltran sees this result as an opportunity to better understand this kind of cancer.
"We don't necessarily know at this time how to subcategorize all the variants in between adeno and neuroendocrine [prostate cancer]," she said. "What we try to highlight is that this phenotypic score can subgroup patients based beyond histology itself into different variants within that spectrum."
Chun-Chieh Lin, a pathologist with Dartmouth University who was not involved in the study, praised NEMO's rigorous validation.
"Overall," he said in an email, "the idea is novel, the data is solid, and the clinical utility in identifying transformation/trans-differentiation of prostate adenocarcinoma into treatment-resistant neuroendocrine prostatic carcinoma is evident."
Lin said that NEMO provides an objective parameter to evaluate the aggregated contributions from a heterogenous tumor population, which is usually clinically challenging in small biopsies. The same rationale used by Beltran and her colleagues, he said, could be applied to other cancer types, potentially enabling applications in early detection, monitoring treatment response, and detecting minimal residual disease.
"However," he added, "the method's detection limit is not fully exploited, as the specimens with tumor content under 3 percent [were] excluded. As the majority of prostatic adenocarcinoma show some degree of neuroendocrine differentiation, this will be of clinical importance in monitoring the non-metastatic lesions, which are characterized by low cell-free DNA [and] thus low tumor content in the plasma."
Lin also noted that the method relies on bisulfite conversion, a process known to have incomplete conversion and to induce extensive DNA damage, which might affect sequencing results, but could be mitigated by recent advances in ultrafast bisulfite sequencing.
Beltran agreed that more studies are needed to further refine NEMO and said some are in the works.
Although it wasn't the main focus of the current study, for example, her group also evaluated NEMO's performance in the lung cancer setting, where it discriminated small cell lung cancer from lung adenocarcinoma. Beltran said that over the coming year, she and her colleagues would be investigating lung and other cancer contexts for possible further applications of NEMO.
Beltran's group has also applied for a patent for NEMO and is interested in possible commercial partnerships to carry NEMO toward the open market.
"We'd love to continue to develop the assay," she said. "We would be interested in any partners that would be interested in developing this with us, but we're still in the early phases of thinking about that."