NEW YORK – Patients with small cell lung cancer could soon benefit from a more personalized approach to treatment than they have seen in years past, thanks to a group of researchers who have defined four subtypes of the disease, each potentially requiring unique treatment strategies.
The researchers, co-led by MD Anderson Cancer Center's Lauren Byers, detailed these four subtypes in a paper published in the journal Cancer Cell last week. The subtypes, dubbed SCLC-A, SCLC-N, SCLC-P, and SCLC-I, are rooted largely in their differentiated transcription factors. SCLC-A is characterized by activation of the ASCL1 gene, SCLC-N of the NEUROD1 gene, and SCLC-P of the POU2F3 gene. SCLC-I, in contrast, is characterized by low expression of ASCL1, NEUROD1, and POU2F3, but carries an inflamed immune signature comprising high expression of numerous immune checkpoint molecules.
Byers highlighted the outsized discrepancy in progress between personalized treatments for non-small cell lung cancer versus SCLC and noted that these four subtypes could constitute a major step toward bringing precision oncology advances to SCLC.
"We now have a path forward to say for the first time that we can start defining the subtypes of small cell and not have to treat it as a single entity," she said. "I really think this classification system will be the beginning of doing what EGFR mutations and ALK fusions did for non-small cell [lung cancer] … in terms of developing more effective, more personalized treatments."
Subtype validation, immunotherapy implications
Although previous studies have explored stratifying SCLC tumors based on ASCL1, NEUROD1, and POU2F3 gene expression, it was apparent to Byers and her colleagues that these subtypes would need further validation. As a system of classification, these groups were also incomplete, since a large proportion of SCLC tumors did not fit neatly into any of the three buckets.
"We had a good idea of what some of these subgroups might be, but we said, 'Let's let the data tell us,'" Byers said. To accomplish this, she and her colleagues used a cohort of 81 early-stage SCLC tumors, for which gene expression information was available, and ran the RNA sequencing data through a computational model called non-negative matrix factorization.
While the subgroups that emerged from the data were mostly expected and in line with prior groupings, the fourth subgroup — which the researchers are calling SCLC-I — came as a surprise. "That [SCLC-I group] was really exciting because a big clinical question for all of us has been, 'What's going to be the biomarker for patients that respond to immunotherapy?' When we saw that inflamed signature come out of the initial analysis, we predicted that was probably the group of patients that are most likely to benefit from immunotherapy," said Byers.
This prediction was rooted in the finding that the inflamed signature identified in the SCLC-I subgroup consisted of higher expression of CD174, which encodes PD-L1; PDCD1, which encodes PD-1; CD80 and CD86, which encode CTLA4; and countless other immune checkpoint molecules. As such, researchers expect that patients with tumors characterized by this signature are more likely to respond to immune checkpoint inhibitors, such as anti-PD-L1, anti-PD-1, and anti-CTLA4 agents.
The next step for Byers' team was to validate the four groups and to test their hypothesis about SCLC-I tumors. They used data collected from extensive-stage SCLC patients enrolled in the IMpower133 clinical trial of atezolizumab (Genentech's Tecentriq) plus chemotherapy versus chemo and placebo. As the authors were careful to point out in their Cancer Cell paper, the IMpower133 trial, which led to the first US Food and Drug Administration approval of a frontline immunotherapy for advanced SCLC, was not statistically powered for a subtype-specific analysis.
That said, there were RNAseq data available for the 276 treatment-naïve SCLC patients in the trial, as well as mature outcomes data, making for a useful validation cohort. The fact that the patients in the IMpower133 trial had extensive-stage SCLC was significant as well, since the discovery cohort had only included data from patients with limited-stage SCLC — an uncommon scenario given the aggressive, difficult-to-treat nature of SCLC tumors. Only about one-third of SCLCs are deemed limited stage — meaning they have not yet metastasized widely through the body — upon diagnosis.
Through their IMpower133 validation analyses, Byers and colleagues once again observed the four SCLC subtypes defined by gene expression or the inflamed immune signature. As suspected, they also found a trend toward improved overall survival following atezolizumab plus chemo among patients with the SCLC-I versus the other three subtypes. The same trend was not observed in the comparator arm, where patients received chemo plus placebo, suggesting that the inflamed immune signature characterizing SCLC-I tumors was likely predictive of atezolizumab benefit and not just indicative of a generally better prognosis compared to the other three subtypes. Specifically, in the chemo plus atezolizumab arm of the study, the researchers observed about a 44 percent lower risk of death in the SCLC-I group versus in the other subtypes.
Additional treatment implications
Having discovered potentially significant therapeutic implications for the SCLC-I subtype, the researchers set out to shed light on treatment implications for the other subgroups, as well. In the absence of another large Phase III clinical trial like IMpower133 with available RNA sequencing data, Byers and her team performed this portion of their study in vitro, using cell lines representing each of the four subtypes. They then used data from over 500 different drugs and analyzed the response data according to the SCLC subtype cell lines.
One finding in these cell-based studies with potential clinical implications was that SCLC-P tumors appeared more sensitive to PARP inhibitors compared to other SCLC subtypes. A similar phenomenon was observed for AURK inhibitors in the SCLC-N subgroup and BCL2 inhibitors in the SCLC-A subgroup. The cell-line analysis also revealed that SCLC-I tumors may be more sensitive to the BTK inhibitor ibrutinib (Janssen/AbbVie's Imbruvica) than the other three subtypes.
Finally, when analyzing cell surface proteins, the researchers found that SCLC-A cells had the highest expression of DLL3, which has been identified as a potential target of interest for CAR T-cell therapies and bispecific T-cell engager (BiTE) therapies. SCLC-P and SCLC-I, on the other hand, expressed virtually no DLL3.
Of course, all of these in vitro drug response data will need future validation in vivo — and ultimately in clinical trials — but as initial findings, they represent potential next steps for applying a precision treatment approach to SCLC.
Establishing assays, advancing to the clinic
Indeed, as a next step, Byers explained that she and her colleagues are working on validating the tests used to stratify patients into these four subtypes and then using those tests in clinical trials to definitively zero in on groups of SCLC patients who are more likely to derive benefit from certain treatments. Right now, these tests are immunohistochemistry assays that identify expression of ASCL1, NEUROD1, POU2F3, and the markers that are higher in the inflamed group.
"One of the things we're looking forward to as a next step is further validating those types of markers that can readily be used in the clinical setting," Byers explained, adding that a future goal will be to develop blood-based tests to identify these biomarkers in circulating tumor DNA, as well. For now, however, the IHC assays are likely to provide valuable information alongside clinical trials. One of these tests is now being used to select patients for a trial of PARP inhibitors for SCLC.
"Immunohistochemistry is an immediately available way to start doing this," Byers said. "Maybe in the future there will be other technologies where we can look at larger and larger numbers of biomarkers that will end up becoming standard as we further refine [these subtypes] … but to begin with, we've already shared these immunohistochemistry assays with other colleagues in different places around the world so they can start incorporating them into the research that they're doing."
Intratumoral heterogeneity, subtype switching
Importantly, as an additional element of their study, Byers and colleagues performed single-cell RNA sequencing on SCLC tumors in animal models and found that, in some cases, a single tumor had multiple subtypes. That is to say, it was possible for tumors to be multiple subtypes at once.
"Subtype [intratumoral heterogeneity] can exist, raising the possibility that these subtypes may represent a spectrum or continuum and that subtype ITH … may underlie the natural history of SCLC," the researchers wrote in their Cancer Cell paper. They also noted that it was possible for SCLC-A cells to transition into SCLC-I cells following platinum-based chemotherapy, resulting in potential platinum resistance.
These findings indicated that perhaps the four subtypes were not always mutually exclusive, and given the potential to shift from one to another, it may be prudent to perform biomarker tests throughout the course of treatment rather than just at diagnosis. Byers said it might be possible, since tumor cells might transition to a platinum-resistant subtype after chemo, to mitigate resistance by pairing chemo with a treatment that is active against the new subtype, or blocking tumor cells from shifting in the first place.
"That was something we were really excited about … because these cancers adapt quickly," she said. "At least they're acting in somewhat predictable ways between the subtypes [and] if we know that there are predictable ways that the cancer is going to change to try to evade chemotherapy, then we can anticipate that."
Next steps, future utility
As a conclusion to their Cancer Cell paper, Byers and colleagues suggested that, going forward, it might be possible to perform a single umbrella trial in which patients with SCLC are matched to different personalized treatments based on their subtype classification.
"Subtype could be determined, and then monitored dynamically, by transcriptional, proteomic, or even epigenetic classification," they wrote. "If any one of these predictions demonstrated significant benefit, it would represent the first standard-of-care molecular biomarker selection for SCLC and a foundational step toward personalized therapy for this devastating disease."
While this hypothetical umbrella trial is not yet underway, Byers said that her team has been having discussions with different groups to launch such a study. It will take some time to plan the different biomarker-matched treatment arms in such a trial, but in the meantime, she said, several trials that she and her team are conducting to evaluate new therapeutic approaches or combination treatments for SCLC will incorporate a subtype analysis.
"We are definitely going to be testing the subtypes in all of the patients who are going onto the trials," she said.
"I expect that very quickly, some of the trials that we'll be seeing, not just from our group, but from many of the small cell investigators, will start incorporating [this subtyping system] routinely as one of their biomarkers," she continued. "I'm confident that there will be trials coming out that will start selecting patients based on this in the not too distant future."