NEW YORK – Researchers are beginning to uncover biomarkers to help guide treatment for small cell lung cancer, which to date has not benefited from precision oncology advances.
Small cell lung cancer is an aggressive disease that accounts for about 15 percent of all lung cancers. While patients often initially respond to chemotherapy, the disease typically becomes resistant quickly, and the search for biomarkers that might predict treatment response has been hampered by tissue availability.
"Small cell lung cancer has really been a challenging disease for decades to make any headway on," Erin Schenk, an assistant professor at the University of Colorado Denver's Anschutz Medical Campus, said. "Biomarkers would be tremendously helpful so that we could — even if it's just a small proportion of patients at the start — potentially improve outcomes for these folks."
Unlike non-small cell lung cancer, where patients now have multiple molecularly targeted treatment options, small cell lung cancer patients haven't experienced the same advances. "In the non-small cell lung cancer world — and that's about 80 percent of the patients, roughly — there is a huge panel of biomarkers that I look at clinically before I make treatment decisions," Schenk said.
However, in small cell lung cancer, preclinical work in cell lines and animal models have uncovered heterogeneity that is tied to whether targeted therapies work in those models, Lauren Byers, an associate professor of thoracic/head and neck medical oncology at MD Anderson Cancer Center, said. This suggested to her that biomarkers are going to be important for treating small cell lung cancer.
She and her colleagues have uncovered a biomarker that seems to distinguish small cell lung cancers that do and do not respond to treatment with a PARP inhibitor in combination with chemotherapy. In addition, a Massachusetts General Hospital Cancer Center-led team has found a four-gene signature by using patient-derived xenograft models that similarly distinguishes sensitive and resistant tumors.
"We know that when we treat a handful of small cell lung cancer patients with a certain therapy, we're going to see a range of responses," first author Anna Farago from the MGH Cancer Center said. "And, if we were able to in advance identify those patients better suited for a particular therapy and can direct them down that treatment path as opposed to another one, that would be fantastic."
Searching for markers
Small cell lung cancer has lagged behind non-small cell lung cancer in the search for treatment biomarkers in part due to the lack of available tissue for research. Until very recently, Schenk noted, small cell lung cancer patients had largely been treated the same way for about 40 years — with radiation and chemotherapy. As few patients undergo surgery, there often isn't much tissue available to analyze.
Additionally, when researchers have had tissue samples from initial biopsies or from the few patients who do undergo surgery, they have found a huge number of mutations within small cell lung cancer tumors. However, these mutations haven't been the druggable oncogene drivers like the ones identified in non-small cell lung cancer, MD Anderson's Byers said.
"There are not specific gene alterations that are picked up by next-gen sequencing that tell us, 'OK, these patients will potentially respond to this type of targeted approach,'" she added.
Based on this, she said she expects biomarkers for small cell lung cancer to instead be gene or protein expression based, an area in which studies are also a little behind gene mutation-based ones, she noted.
Searches, though, have begun. Byers and her colleagues previously examined transcriptomic and proteomic differences between small cell lung cancer and non-small cell lung cancer cell lines. Through this, they found that small cell lung cancer overexpresses genes involved in DNA damage response, like PARP, as well as genes involved in cell cycle regulation, as they reported in Cancer Discovery in 2012. They focused their analyses on PARP as, at the time, there were a number of PARP inhibitors in late-stage development that they thought could quickly be tested in clinical trials if their suspicions held out.
Indeed, she and her colleagues conducted a randomized trial of AbbVie's PARP inhibitor veliparib in combination with the oral chemotherapy temozolomide or temozolomide with placebo. As reported in the Journal of Clinical Oncology last year, there was an overall response rate of 39 percent among patients who received both temozolomide and veliparib compared to 14 percent among those who received temozolomide and placebo. However, Byers noted that these patients still had a fairly short response duration.
As this trial was occurring, there was increasing evidence that SLFN11 might serve as a biomarker for PARP inhibition in small-cell lung cancer. SLFN11, she noted, was a largely unknown protein, but one that is involved in cell death following DNA damage. "It kind of makes sense that, if the PARP inhibitor is causing DNA damage, and they have this SHLFN11 protein, then those are the cells that are going to be killed," she said.
Byers and her colleagues retrospectively examined whether patients with SLFN11-positive tumors fared better on veliparib plus temozolomide. They found those patients had longer progression free and overall survival than those given temozolomide and placebo.
In their new study, MGH Cancer Center's Farago and her colleagues built on that work. They likewise examined a combination therapy for small cell lung cancer that included the PARP inhibitor olaparib (AstraZeneca's Lynparza) and temozolomide. In their single-arm phase I/II study of 50 patients, which was published last month in Cancer Discovery, the researchers reported an overall response rate of 41.7 percent, similar to what Byers and her colleagues had reported.
"It's not a randomized study, but it's certainly a study that shows a really encouraging signal of activity," Farago said.
They additionally reported a median progression-free survival of 4.2 months and a median overall survival of 8.5 months.
As part of their study, Farago and her colleagues conducted what they called a "co-clinical trial," in which they analyzed 32 patient-derived xenograft mouse models' response to treatment.
"Our initial goal with this was just to identify … the molecular features of the tumors that do respond and the features of the tumors that do not respond in a way that takes into account the heterogeneity of the disease," senior author Ben Drapkin, also from MGH Cancer Center, said.
Some of the patient-derived xenograft models they analyzed were from patients who participated in their trial, and for two patients, they developed models from both before and after relapse. Having clinical correlates and these pre- and post-relapse models, Drapkin said, enabled them to better interpret what the mouse models' responses to olaparib and temozolomide meant.
They used this set of PDX mice to search for a signature-predicting response to olaparib and temozolomide to find a set of four inflammatory response genes — CEACM1, TNFSF10, TGIF1, and OAS1 — whose expression distinguished sensitive and resistant models.
Farago and Drapkin and their team also examined Byers' SHLFN11 response biomarker to find in their discovery set that it could distinguish the most sensitive of the mouse models.
Byers said she suspects that the molecular signature uncovered by Farago, Drapkin, and their team — which she said looks "promising as a potential biomarker" — might be another means of identifying the same group of sensitive tumors that her SHLFN11 biomarker does. Farago and Drapkin's signature, she noted, combines genes that are involved in inflammation and interferon response, and SHLFN11 is an interferon-regulated gene.
Byers added that their use of PDX models is "a really elegant way of looking in even more detail at what potential biomarkers might be."
That approach can circumvent to a certain degree the lack of tissue availability to search for biomarkers in small cell lung cancer. These models, she said, allowed them to do more comprehensive profiling that they might not have been able to do with just the small amount tissue that is available from tumor biopsies.
However, PDX models, because of how they are developed by engrafting patient tissue into the mouse model, lack an immune system, which Pierre Massion, a pulmonologist at Vanderbilt University Medical Center, said is a general limitation of PDX models.
In this case, Byers said that preclinical work and work in breast cancer has indicated that immune response to PARP inhibitors could be a key part of why and for how long patients respond to the drugs. But in these PDX models, Colorado's Schenk noted, that response is lost, as there are no immune cells infiltrating the tumors.
"I think one thing that we don't know yet — and that would be really interesting to know — is what the differences are in terms of the immune response between responders and non-responders," added Byers. That, she said, would be an avenue to pursue further.
For the MGH team, Farago said that their molecular signature is multiple steps away from being ready to be used in the clinic. According to Drapkin, one of their next steps is to try to tease out using the same mouse models what is underlying the signature, as the signature includes genes involved in immune response, but the models themselves lack an immune system.
Still, they said it was exciting to see such a strong signal.
"Ultimately, I think if we had a way to clinically subdivide patients based on biomarkers in small cell lung cancer samples — much like the way that we do in non-small cell lung cancer — this could be practice changing, not specifically for our treatment, but more generally," Farago said. "I think that's the direction where we would like to see the field go."