Neoadjuvant Atezolizumab Shows Pathologic Responses, Induces Immune Cell Changes in Resectable NSCLC
Neoadjuvant treatment with the PD-L1 inhibitor atezolizumab yielded a promising pathologic response rate in a phase II study of patients with resectable non–small cell lung cancer (NSCLC). The treatment resulted in significant changes in the immune subset, with an increase in PD-L1–positive tumor-infiltrating immune cells (Abstract 99).
The LCMC3 trial included patients with stage IB, II, IIIA, and selected IIIB resectable and untreated NSCLC. Filiz Oezkan, MD, of The Ohio State University, presented preliminary results. The first 54 patients received neoadjuvant atezolizumab and underwent a presurgical evaluation for major pathologic response (MPR), followed by standard-of-care chemotherapy and subsequent adjuvant atezolizumab.
The median age of those first 54 patients was 65, and 54% were male. Patients had an ECOG performance status of either 0 (65%) or 1 (35%). Most patients were either current (32%) or former (61%) smokers and had nonsquamous (65%) and squamous (35%) histology. A total of 30 patients (56%) had PD-L1 expression of ≥ 1%, while 22% had < 1% PD-L1 expression, and 22% had unknown PD-L1 status.
Most of the patients (96%) completed two cycles of neoadjuvant atezolizumab; the two patients who did not failed to complete a second cycle due to adverse events (AEs), with one case of grade 1 pyrexia and one case of grade 2 dyspnea. Four patients did not undergo surgery (7%), and nine patients delayed surgery (17%).
Among the 45 patients who were evaluated for pathologic regression, there were 10 patients (22%) with an MPR, defined as having ≤ 10% viable tumor cells; three of those patients had a pathologic complete response. Almost all evaluated patients showed some degree of pathologic regression of their tumor. There were three partial responses by RECIST criteria (7%). Forty-two patients (93%) had stable disease, and no patients had progressive disease.
All 54 patients were included in the safety evaluation. Fifty-one patients (94%) experienced at least one all-cause AE; 28% had a grade 3/4 AE, and one patient experienced a grade 5 AE (sudden death not related to study treatment, occurring approximately 2 weeks after surgery). Most patients (59%) experienced a treatment-related AE (TRAE); however, grade 3/4 TRAEs were rare (6%). Sixteen patients (30%) experienced a serious AE.
The most common TRAEs included fatigue (22%), arthralgia (11%), and increased AST and pyrexia (both 9%). Other TRAEs included diarrhea, nausea, increased ALT, decreased appetite, dyspnea, and infusion-related reactions.
Thirty-one patients were available for a biomarker analysis based on paired peripheral blood samples from before and after atezolizumab treatment, analyzed by immune cell phenotyping. This analysis showed significant increases from baseline in CD8+ T cells (p < 0.05), mature natural killer (NK) cells (p < 0.01), late-activated CD16+/CD56+ NK cells (p < 0.05), CD16+ NK cells (p < 0.05), and Th1 response-related dendritic cells (p < 0.05). There was also a significant decrease seen in B cells (p < 0.01), and all these changes were significant both in terms of the relative change and the absolute change from baseline.
In that same subset of patients, those who did not achieve an MPR showed significant increases in late-activated NK cells, a monocytic myeloid cell subpopulation, and a Th2- and Th17-response–related dendritic cell population.
In a separate subset of 31 patients who had paired tumor samples from the time of screening and of surgery, the researchers performed immunohistochemistry staining on the samples; PD-L1 status was available in 23 of those patients, and 16 of those were PD-L1–positive (PD-L1 ≥ 1%). “We observed that in most of the patients, the [tumor-infiltrating] immune cells that express PD-L1 increased after treatment,” Dr. Oezkan said. There was no correlation with PD-L1–positive tumor cells, however.
A final biomarker analysis included evaluation of genetic mutations in a group of 25 patients. The most commonly mutated genes were TP53 (64%) and KRAS (28%). Mutations to ATM, MGA, and STK11 were seen in 20% of patients, respectively.
“LCMC3 biomarker analyses will improve the understanding of lung cancer immunology in the setting of neoadjuvant anti–PD-L1 treatment with atezolizumab,” Dr. Oezkan concluded.
During a panel discussion following the presentation, Roy S. Herbst, MD, PhD, FASCO, of the Yale Cancer Center, who co-chaired the session, called the study a “tour de force,” given the difficulty often seen in accruing patients to neoadjuvant trials. The trial will eventually include 180 patients, of which Dr. Oezkan said 135 have been enrolled to date.
– David Levitan
T-cell Receptor Diversity on an RNA Basis May Help Predict Efficacy of the Immune Response to Premalignancy
A new study has offered the first characterization of the T-cell receptor (TCR) repertoire bronchial premalignant lesions (PMLs), which are precursors of lung squamous cell carcinoma. The study found that TCR diversity could help predict the immune response to premalignancy, but that diversity is not necessarily associated with mutational burden (Abstract 102).
“Antigen recognition is HLA-dependent and TCR-specific,” said Asaf Maoz, MD, of Boston University School of Medicine and Boston Medical Center, who presented results of the study during the ASCO-SITC Clinical Immuno-Oncology Symposium. “Bulk RNA-Sequencing data can be leveraged to characterize the T-cell repertoire.”
The investigators focused on squamous cell carcinoma, which is thought to arise from a spectrum ranging from normal bronchial tissue through squamous metaplasia, mild through moderate and then severe dysplasia, and finally with carcinoma in situ. “Potentially, this is a stage of premalignancy where we can intervene, but before we can do that, we need to know the immune response to these lesions,” Dr. Maoz said.
The investigators used RNA-Seq data from 50 patients who underwent yearly lung cancer screenings with auto-fluorescence bronchoscopy and computed tomography and who were included in the Pre-Cancer Genome Atlas; they underwent a total of 294 endobronchial biopsies and 137 brushings. The mean age of patients at baseline was 58.8 years, 54.4% were male, and 88.0% were white; 96.0% of the cohort were ever-smokers, and 66.0% had a prior history of lung cancer. Furthermore, 56% had chronic obstructive pulmonary disorder, and Dr. Maoz said these were patients at very high risk for lung cancer.
Using the TCR Repertoire Utilities for Solid Tissue/Tumor (TRUST) tool, the analysis found 40,421 unique TCR sequences ( > 5 amino acids). Of those, 3,396 (8.4%) were found in more than one sample, and 1,057 (2.6%) were found in two or more patients; those found in two or more patients were considered “public,” while those found in fewer were deemed “private” TCRs.
They then compared these sequences to those included in curated databases of known antigen specificities; Dr. Maoz pointed out that those databases are highly biased toward infectious antigens rather than tumor antigens. Despite of that caveat, the public TCRs were more likely to have a known target; this was seen in 53 of 1,057 TCR sequences (5%), compared with 334 of 39,364 of private TCRs (0.85%; p < 0.0001).
The diversity of TCRs was found to be negatively associated with a transcriptional signature of T-cell mediated immune activation (p < 0.001). There was no correlation, however, between that diversity and mutational burden.
The investigators also examined how TCR diversity was associated with progression among samples displaying a proliferative molecular subtype; this subtype is associated with dysplasia and the risk of disease progression. They found that TCR diversity decreased in PMLs that regressed, compared to the PMLs that progressed or were stable (p = 0.045).
Dr. Maoz pointed out several limitations to the analysis, including that bulk RNA-Seq can capture only a fraction of the T-cell receptors and the lack of functional data. There was also limited mutational data available, meaning it was difficult to decipher what antigens are driving an immune response.
“Despite these limitations, we think this is a very important characterization of the TCR repertoire,” he said. “We showed that T-cell receptor diversity on an RNA basis may help predict the efficacy of the immune response to premalignancy.”
Steven M. Dubinett, MD, of the David Geffen School of Medicine at the University of California, Los Angeles, discussed the study. “This is an important abstract and topic, in that squamous cell lung cancer presents new challenges in understanding premalignancy,” he said. He noted that only a subset of premalignant lesions progress to become an invasive tumor while many will regress without intervention, and that further complicating their management is the fact that at least half the time a PML is observed, a subsequent lung cancer is observed at a different site.
“It is useful that bulk RNA-Seq data can be leveraged to create a partial picture of the TCR repertoire in premalignancy,” Dr. Dubinett said. It remains to be seen how a targeted RNA-Seq approach might differ from this method, as do the relationships or correlations between these results and the tumor microbiome. He also noted that characterizing PMLs in pulmonary adenocarcinoma, which is more common than squamous cell carcinoma, may present even more challenges.
“Investigations are beginning to reveal the role of the immune response in regulating the course of events in the very earliest stages of pulmonary premalignancy,” Dr. Dubinett said. “These studies are anticipated to facilitate the eventual application of lung cancer interception, which achieves the prevention of invasive disease.”
– David Levitan
Plasma ctDNA Targeted Sequencing Could Help Predict Immunotherapy Response in NSCLC
A targeted gene sequencing approach covering a limited number of relevant genetic alterations using plasma circulating tumor DNA (ctDNA) could help predict treatment responses to PD-1 blockade in advanced non-small cell lung cancer (NSCLC), according to a new study (Abstract 103). This could represent a noninvasive, relatively simple approach to help guide treatment modifications, but it still requires prospective validation.
Tumor mutational burden (TMB) is an emerging biomarker for immunotherapy response, but it has a number of limitations, according to Nicolas Guibert, MD, PhD, of the University Hospital of Toulouse, in France, who presented the new study during the ASCO-SITC Clinical Immuno-Oncology Symposium. Assessing TMB involves expensive bioinformatics, and has limited feasibility due to the availability of suitable tissue and the need for whole-exome sequencing or at least very large gene panels.
“We hypothesized that by using some more limited gene panels and doing targeted sequencing of ctDNA, focusing on the limited numbers of relevant alterations, we could predict response to PD-1 inhibitors,” Dr. Guibert said.
The investigators used next-generation sequencing (NGS) focusing on alterations to 36 genes in a cohort of 98 patients who received nivolumab in the second-line setting for advanced NSCLC. The sequencing was done on plasma samples collected before nivolumab initiation and 1 month after treatment began. Of 86 evaluable patients, 39 responded to anti-PD-1 therapy and 47 did not. A total of 67 of the evaluable patients had detectable alterations in ctDNA, and 19 had none.
They found several differences in outcomes based on the presence or absence of certain alterations detected in ctDNA. Among patients with a PTEN or STK11 alteration (14 patients), the median progression-free survival (PFS) was 2 months; among those with no such alteration, the median PFS was 8 months.
The opposite was seen with KRAS or TP53 mutations; among those with one or both of these mutations (34 patients), the median PFS was 11 months, while those with neither mutation had a median PFS of only 2 months.
Based on these results, the researchers proposed an algorithm that could be used to select patients for immunotherapy based on a ctDNA profile. The algorithm is based on the presence of ctDNA, targetable drivers including BRAF V600E, EGFR, or ALK (which would lead to treatment with genotype-directed targeted therapy), and then STK11/PTEN and KRAS/TP53 mutations. Those without detectable ctDNA had a median PFS of 3.5 months in this cohort; those with an STK11 or PTEN mutation were deemed “immune low” and had a median PFS of 2 months, and should likely be treated with chemotherapy. Those with the KRAS and/or TP53 mutations are considered “immune high,” and had a median PFS of 14 months, and should be treated with PD-1 inhibitors.
The study also found that early variations in ctDNA can accurately discriminate responders from nonresponders to anti-PD-1 therapy. Patients who had a decrease at 1 month had a median PFS of 10 months, and a 6-month PFS rate of 74%; those who saw an increase in ctDNA at 1 month had a median PFS of 2 months, and a 6-month PFS rate of only 16%. Dr. Guibert noted that changing the cutoff point for allelic fractions changes this predictive effect.
“Plasma NGS covering a limited number of relevant alterations can accurately predict sustained responses under PD-1 blockade,” Dr. Guibert concluded. “Using one single test, plasma NGS can noninvasively guide patients to receive either targeted therapy or immune checkpoint inhibitors and follow response to treatment.”
Steven M. Dubinett, MD, of the David Geffen School of Medicine at the University of California, Los Angeles, discussed the study and pointed out several outstanding questions for this approach. “Will the performance characteristics of these plasma ctDNA tests be adequate to alter therapy at early time points?” he asked. He also wondered if the targeted sequencing approach may be adequate in NSCLC; in this study, it was difficult to discern what was driving some of the outcomes, such as whether the limited mutational burden seen with EGFR or the presence of the STK11 mutation was responsible for the poor outcome in some patients.
Dr. Guibert also noted that this was a retrospective study with a limited number of patients, and the approach still needs to be prospectively validated in larger and unselected patient populations. – David Levitan
– David Levitan
Anti–TIM-3 Antibody LY3321367 Well Tolerated, Appears Promising in Phase I Trial
The anti–TIM-3 antibody LY3321367 was well tolerated both as monotherapy and in combination with an anti–PD-L1 antibody known as LY3300054, according to results of a phase Ia/Ib trial of patients with advanced solid tumors (Abstract 12). The TIM-3–targeted agent showed some preliminary antitumor activity in the monotherapy setting.
“TIM-3, when activated, leads to suppressive signals to T cells and seems to be a hallmark of T-cell exhaustion in both malignancy as well as infection,” James J. Harding, MD, of the Memorial Sloan Kettering Cancer Center and Weill Cornell Medical Center, said during his presentation of trial results. “When you block TIM-3…you can reinvigorate T-cell function.” TIM-3 is expressed on several cell types, including T cells, natural killer cells, and macrophages; it may be associated with mediation of intrinsic and acquired resistance to PD-1 blockade.
The phase Ia dose-escalation study included two arms: in arm A, patients received TIM-3–targeted monotherapy with LY3321367 (23 patients), and in arm B, patients received that agent along with the PD-L1–targeted LY3300054 (18 patients). All patients had advanced solid tumors, had progressed on standard therapies, had an ECOG performance status of 0 or 1, and had adequate organ function.
In the monotherapy group, 56.5% of patients were female, and the median age was 54.0 years. Six patients had lung cancer (26.1%), three had breast cancer (13.0%), two had colorectal cancer (8.7%), and 12 patients had other types of malignancy (52.2%). In the combination arm, 44.4% of patients were female, and the median age was 65.0 years. Three patients had colorectal cancer (16.7%), two had bladder cancer (11.1%), one each had breast and small cell lung cancer (5.6% and 5.6%), and 11 patients had other types of malignancy (61.1%). In both study arms, a majority of patients had received three or more prior lines of therapy; in arm A, 34.8% had received anti–PD-L1 therapy, while 16.7% had in arm B.
In both treatment groups, there were no dose-limiting toxicities, dose-limiting equivalent toxicities, or treatment-related adverse events (TRAEs) of grade 3 or higher. In the monotherapy group, maculopapular rash occurred in three patients (13.0%) at the 70 mg, 600 mg, and 1,200 mg dose levels (this dose level received 600 mg from cycle three onward). Pruritus occurred in two patients (8.7%) at the 200 mg and 600 mg dose levels. Other AEs that occurred in only one patient included abdominal pain, arthralgia, fatigue, and nausea. The dermatologic toxicities were deemed to be immune-related.
In the combination cohort, four patients had fatigue (22.2%) at dose levels of 70 mg, 200 mg, and 600 mg. Three patients each reported decreased appetite and infusion-related reactions (16.7%) at the 70 mg and 600 mg doses. Maculopapular rash occurred in two patients receiving the 70 mg dose, while other TRAEs occurring in only one patient included abdominal pain, breast pain, diarrhea, and others.
Dr. Harding said that a high percentage of patients had treatment-emergent antidrug antibodies (ADA) in both the monotherapy and combination therapy groups. However, most patients had low ADA titres, and there was no interference with pharmacokinetics or exposure.
Preliminary efficacy results were available only for the monotherapy cohort. Ten patients had stable disease as their best response, and there was one partial response, yielding a disease control rate of 47% (11 out of 23 patients). The patient with a partial response was a 54-year-old male with extensive small cell lung carcinoma, who had progressed on cisplatin/etoposide, radiotherapy, and ipilimumab/nivolumab.
Dr. Harding said the recommended phase II dose was set at 1,200 mg infusions every 2 weeks for the first two cycles, followed by a dose of 600 mg from cycle three onward. He noted that expansion cohorts with both the single-agent TIM-3–targeted antibody and the combination are ongoing in various malignancies, including lung cancer, gastric cancer, hepatocellular carcinoma, head and neck cancer, and bladder cancer.
Michael Morse, MD, of Duke University Medical Center, was the discussant for the study. He called the new study “an example of the direction we’re moving in the field, trying to expand therapies for patients with more difficult-to-treat cancers.”
Dr. Morse noted that the efficacy data is too preliminary to draw firm conclusions, and it is important to remember that the cancers studied in the trial included malignancies that generally are less responsive to immune checkpoint blockade. “TIM-3 is a very complicated target,” he said. “It may be working in different ways, and we really do need to get data from the tissue level to understand this better.” He added that “we eagerly await the combination data.”
– David Levitan
HER2-Targeted Vaccine Improves DFS in Some HER2 Low–Expressing Patients With Breast Cancer
Combining the HER2-targeted vaccine nelipepimut-S (NPS) with trastuzumab and granulocyte-macrophage colony-stimulating factor (GM-CSF) showed a trend toward improved disease-free survival (DFS) compared with trastuzumab and GM-CSF alone in patients with high-risk, HER2 low–expressing breast cancer (Abstract 1). There was a significant improvement seen specifically in those with triple-negative breast cancer (TNBC).
Most patients with breast cancer express HER2, but only those with higher levels of expression are eligible to receive targeted therapy, such as trastuzumab. “The bulk of these patients have a lower level of HER2 expression and do not benefit from HER2-targeted therapy,” G. Travis Clifton, MD, of the San Antonio Military Medical Center, said during his presentation of the results of the phase IIb trial. This covers approximately 60% to 70% of patients. “Because these patients’ tumors do have HER2 expression, we may be able to target those with vaccination that targets the HER2.” Preclinical and translational data strongly suggest that there may be a synergistic effect between trastuzumab and a HER2-targeting CD8+ T-cell–eliciting vaccine.
NPS is such a vaccine—derived from the extracellular domain and designed to stimulate CD8+ T cells. The new study included 275 women with high-risk invasive breast cancer with HER2 expression of 1 to 2+, as measured by immunohistochemistry. All patients were clinically disease free after receiving standard-of-care therapy. They were randomly assigned to receive either NPS along with trastuzumab and GM-CSF (136 patients) or trastuzumab and GM-CSF alone (139 patients); the primary endpoint was DFS at 24 months.
The treatment groups were well balanced in terms of demographic characteristics. The median age in the vaccine group was 52.2 years, compared with 50.5 years in the control group, and most of the total patient population was white. In the NPS group, 43% had received adjuvant chemotherapy and 53% had received neoadjuvant chemotherapy; in the control group, those rates were 41% and 55%, respectively.
Most of the intervention cohort (94.3%) experienced at least one treatment-related adverse event (TRAE), but there was no added toxicity with the vaccine (p = 0.17 between groups). The majority of TRAEs in both groups were grade 1 or 2, primarily consisting of local injection site reactions, skin induration, pruritus, and fatigue. There was no difference between the groups with regard to cardiac ejection fraction over time (p = 0.65) or at each measured time point. There was no additional cardiac toxicity with the addition of NPS.
The trial was stopped at an interim analysis, partially because it was determined that additional follow-up was unlikely to change the results and, also, the results of the phase III NSABP B-47 trial of in a similar patient population showed that the addition of trastuzumab to standard adjuvant chemotherapy does not benefit patients with HER2 low–expressing tumors.1 This was the treatment for the control arm in this trial, so the decision was made that it should be stopped.
After a median follow-up of 24.7 months, there were a total of 12 recurrences in the NPS group (8.8%), compared with 20 recurrences in the trastuzumab and GM-CSF–alone group (14.4%). The 24-month DFS rate was 89.8% with the vaccine compared with 83.8% without it (HR 0.62, 95% CI [0.31, 1.25]; p = 0.18). At 36 months, the DFS rate was 86.7% with NPS and 80.8% without it.
In only patients with HR-positive breast cancer, the 24-month DFS rate was 87.9% with NPS and 90.9% without it (HR 1.19, 95% CI [0.46, 3.01]; p = 0.71). This DFS rate is comparable to the NSABP B-47 trial.
In contrast, the vaccine did result in benefit for patients with TNBC. In this group, the 24-month DFS was 92.6% in 53 patients who received NPS, compared with 70.2% in 44 patients who did not (HR 0.26, 95% CI [0.08, 0.81]; p = 0.013). The 36-month DFS rate in the NPS group was 82.3% compared with 70.2% in the control group.
“This is a subset of a trial that was stopped at an interim analysis, but we think the results are intriguing,” Dr. Clifton said. “It may provide benefit in patients with HER2 low–expressing TNBC, but this needs to be confirmed in a [larger] trial,” which he said is now being pursued.
Kunle Odunsi, MD, PhD, of the Roswell Park Comprehensive Cancer Center, was the discussant for the study. He noted that HER2 low–expressing breast cancer represents a very large patient population, so “if this [approach] is positive, potentially you could be looking at helping a large base of breast cancer patients.”
Dr. Odunsi said that the mechanistic rationale for the NPS vaccine is sound. The trial’s strengths, he said, include the preclinical demonstration of in vivo induction of peptide-specific cytotoxic T-lymphocyte response, and the result specifically in patients with TNBC “appears to be clinically meaningful.”
Some questions do remain, however, including whether the study had adequate power for the subgroup analysis of TNBC. “This is an immunoprevention trial, where you are trying to do a secondary prevention, and therefore the T cells you are trying to generate ideally should have memory attributes,” Dr. Odunsi said. “The ability to have secondary recall response is going to be critical.”
– David Levitan
IL-12 Plasmid Injections May Add Benefit to Neoadjuvant Chemotherapy for Ovarian Cancer
The addition of the formulated interleukin (IL)-12 plasmid GEN-1, delivered intraperitoneally, to doublet chemotherapy was safe and appeared to have antitumor activity in a phase I trial of patients with ovarian, fallopian tube, or primary peritoneal cancer (EOC) undergoing neoadjuvant therapy (Abstract 2). The treatment appeared to change the tumor microenvironment, and further studies are ongoing.
Both surgery and chemotherapy are important parts of the treatment paradigm for ovarian cancer, Premal H. Thaker, MD, of the Washington University in St. Louis School of Medicine, said during her presentation of the study results. “However, in the past 5 years…there has been a raise in terms of neoadjuvant treatment, since 80% of our patients will relapse even with this paradigm of treatment.” Ovarian cancer has long been thought to be an immunogenic malignancy. “We need newer immunomodulatory treatment strategies to help improve the treatment of these patients,” she said.
IL-12 has several mechanisms of action, including an ability to promote adaptive immunity with T-cell proliferation and activation, as well as promoting innate immunity through activation and proliferation of natural killer cells. Dr. Thaker said it induces a shift of the tumor microenvironment from immune suppression toward immune activation.
GEN-1 is an IL-12 plasmid formulated with a lipopolymer, encoding the p35 and p40 subunits of the IL-12 gene. It is designed for intraperitoneal injection and can produce durable local levels of IL-12 and other related cytokines after a single injection. The investigators hypothesized that when added to standard doublet chemotherapy, GEN-1 would stimulate a potent immune response in patients with ovarian cancer, resulting in improved R0 resection rates, reduced immunosuppression in the tumor microenvironment, and enhanced T-cell antitumor activity.
The phase I study included 18 patients in the intent-to-treat (ITT) population and 14 in the per-protocol population. All patients had newly diagnosed EOC and received paclitaxel and carboplatin along with eight weekly cycles of GEN-1; the study used a dose-escalation design that raised the GEN-1 intraperitoneal dose in approximately 30% increments.
The median age of the ITT population was 63 years, and 95% had serous histology; 67% had stage IIIC disease and 33% had stage IV disease.
The most common adverse events (AEs) attributed to GEN-1 included nausea (60%), abdominal pain/cramping (40%), fatigue (40%), and vomiting (40%). Most AEs were grade 1 or 2; there were two cases of grade 3 diarrhea (13%) and one case each (6%) of grade 3 abdominal pain/cramping, vomiting, and neutropenia. There was one grade 4 case of neutropenia. Four patients discontinued the study due to AEs: one had dosing delays of more than 21 days, one had declining performance status, one had sepsis and congestive heart failure, and one had altered taste related to GEN-1. There were no dose-limiting toxicities, and the maximum tolerated dose was not reached in this study.
Among the 14 patients in the per-protocol population, there were two complete responses based on RECIST criteria and 10 partial responses; two patients had stable disease. All 14 underwent interval debulking surgery, and nine patients had R0 debulking, three patients had R1 debulking, and two patients had R2.
The median progression-free survival (PFS) for the full ITT cohort was 17.1 months, and 21.0 months for the per-protocol cohort.
The researchers found changes to the tumor microenvironment after treatment with GEN-1. There were changes to immunosuppressive markers in response to both low and high doses of the treatment; these included FoxP3, IDO-1, PD-1, and PD-L1. There was also an increased ratio of CD8+ T cells to those immunosuppressive markers in 60% to 80% of patients following treatment.
“Adding GEN-1 to doublet treatment is safe and appears to be active in EOC patients receiving neoadjuvant chemotherapy,” Dr. Thaker said. She noted that the OVATION 2 trial is now ongoing; in a phase II portion of that trial, investigators will randomly assign 130 patients with stage III/IV ovarian cancer to receive neoadjuvant chemotherapy alone or along with GEN-1, with PFS as the primary endpoint (NCT03393884).
Kunle Odunsi, MD, PhD, of the Roswell Park Comprehensive Cancer Center, was the discussant for the study. “What is the molecular/cellular response of ovarian cancer to chemo-immunotherapy?” he asked. He said the design of the GEN-1 platform is “very elegant,” in that it combines nanoparticle and gene therapy approaches to influence the tumor microenvironment.
“A major strength of the study is the repetitive biopsies,” he said, noting that the trial involved blood and peritoneal fluid samples at several time points, allowing for a thorough characterization of changes to the tumor microenvironment. Dr. Odunsi posed several important questions for the future, including whether there is potential for combination therapies involving GEN-1 and if there is any “preclinical data that the addition of IL-12 beyond the neoadjuvant setting could be of benefit.”
– David Levitan
Lenvatinib/Pembrolizumab Shows Promise in UC But Managing Toxicity is Challenging
The combination of lenvatinib and pembrolizumab yielded promising antitumor activity and progression-free survival in patients with metastatic urothelial carcinoma, according to a new phase Ib/II study (Abstract 11). However, some concerns remain regarding the toxicity of the combination.
Urothelial cancer accounts for more than 90% of all bladder cancers. Pembrolizumab monotherapy is currently approved as a first-line therapy for patients with urothelial carcinoma who are ineligible to receive cisplatin or other platinum-based chemotherapy. It is also approved as second-line treatment for patients with advanced/metastatic urothelial cancer. Lenvatinib, a multikinase inhibitor of VEGFR 1-3, FGFR 1-4, PDGFRα, RET, and KIT, is used as single-agent therapy in several malignancies, including thyroid cancer and hepatocellular carcinoma.
“Lenvatinib combined with a PD-1 monoclonal antibody has enhanced antitumor activity in preclinical models,” said Nicholas Vogelzang, MD, FACP, FASCO, of The US Oncology Network, McKesson Specialty Health, who presented the results of the new study.
Dr. Vogelzang discussed the phase II portion of the study, which included 20 patients with histologically confirmed metastatic urothelial carcinoma. Patients had received no more than two prior systemic regimens and had an ECOG performance status of 0 or 1. They received oral lenvatinib 20 mg once daily along with intravenous pembrolizumab 200 mg administered once every 3 weeks.
The median age of patients enrolled was 72, and 70% of the cohort was male. Most patients had an ECOG performance status of 1 (70%). Half the cohort were PD-L1 positive, 40% were PD-L1 negative, and two patients (10%) did not have their PD-L1 status assessed. All 16 patients who had prior therapy received platinum-based chemotherapy.
The primary outcome in the trial was immune-related RECIST (irRECIST) objective response rate (ORR) at 24 weeks; five patients achieved such a response, for an ORR at 24 weeks of 25%. The total ORR was also 25%, meaning no further responses were seen after 24 weeks. The result did not change when modified RECIST (mRECIST) criteria were used.
One patient had a complete response, and four patients (20%) had partial responses. Another nine patients (45%) had stable disease, yielding a clinical benefit rate of 40%. Again, the results were the same using mRECIST criteria.
“Although there were five objective responses, there were an additional seven patients or more who had minor regressions of disease,” Dr. Vogelzang said. “Clearly, [it was] an active regimen.”
At the time of data cutoff, three patients were still undergoing the study treatment. Two of these were PD-L1 negative, and one was PD-L1 positive. One patient withdrew consent but remained in response at data cutoff.
After a median follow-up of 11.7 months, the median PFS by both irRECIST and mRECIST was 5.4 months. At 3 months, the PFS rate was 60%; at 6, 9, and 12 months, the PFS rate was 35%, 26%, and 26%, respectively.
Eighteen of the 20 patients (90%) experienced treatment-related adverse event (TRAE) of any grade, and 50% of the total patient cohort had a grade 3 or 4 TRAE. Six patients (30%) had a serious TRAE, and there was one death from gastrointestinal hemorrhage. Dr. Vogelzang said it was more likely related to lenvatinib than pembrolizumab, although it was impossible to be certain. Most of the cohort (75%) had a TRAE that led to a study drug dose adjustment, including four patients (20%) who withdrew/discontinued therapy, seven patients (35%) who required a dose reduction, and 12 patients (60%) who required a dose interruption.
The most common TRAEs of any grade included proteinuria (45%), diarrhea (40%), and hypertension (35%). Grade 3 or 4 proteinuria, diarrhea, fatigue, and increased lipase were each observed in two patients (10%).
“Lenvatinib plus pembrolizumab demonstrated promising antitumor activity with manageable adverse events,” Dr. Vogelzang said. “This response rate warrants further investigation, and the lenvatinib plus pembrolizumab combination will be studied in a phase III trial in urothelial carcinoma.”
During a panel discussion, several questions focused on the toxicity of the combination. “Is this a viable strategy for metastatic urothelial carcinoma?” asked Robert Dreicer, MD, MS, MACP, FASCO, of the University of Virginia Cancer Center, who co-chaired the session. “The toxicity was sort of hair-raising.”
Dr. Vogelzang acknowledged that the dose of lenvatinib used was higher than that used in other malignancies, and that many clinicians will begin the agent at a lower dose and then raise it as tolerated. He said that as combination trials with immunotherapy agents along with other drugs proliferate, it becomes important to demonstrate bigger survival advantages to justify toxicity that can double, or even triple, compared to checkpoint inhibitor monotherapy.
William Y. Kim, MD, of the University of North Carolina Lineberger Comprehensive Cancer Center, who co-chaired the session with Dr. Dreicer, said that “cisplatin in particular has an associated decrease in renal function,” and many combination trials focus on previously treated patients. Inclusion criteria on those trials that try to reduce some toxicity by setting cutoffs for renal function will make enrollment on the trials increasingly difficult, he said.
– David Levitan
Plasma Proteomic Expression Correlated With Immunotherapy Response
Plasma proteomic profiling of patients with melanoma treated with anti–PD-1 therapy found a number of proteins are differentially expressed between responders to the therapy and nonresponders (Abstract 130). Further validation could allow this form of liquid biopsy to help guide treatment.
Several noninvasive biomarkers to aid in predicting response to immunotherapy are under investigation, including circulating tumor cells, cell-free DNA or RNA, and exosomes, but the best predictors of response so far involve tissue biopsy for immune profiling. “To date, no single noninvasive approach has proven particularly effective in building a good predictor of therapy response, and, particularly, response to immunotherapy,” said Arnav Mehta, MD, PhD, of Massachusetts General Hospital, who presented results of the study. “It is likely the most effective noninvasive predictors will involve layered datasets, utilizing multiple sources of information.”
The investigators examined whether plasma proteomic profiling might represent one of those layered datasets. They partnered with Olink Proteomics and used a proximity extension assay that allowed for the quantification of expression of proteins; this was done using 10 to 20 µL of plasma. The cohort included 58 patients with melanoma who were treated with either pembrolizumab or nivolumab. Of the 58 patients included, 44 responded to therapy, and 14 did not. The researchers obtained a total of 149 samples, at baseline, while on treatment, and post-treatment; the on-treatment sample was taken an average of 67 days from the initiation of therapy. They detected a total of 1,104 proteins.
Of the 1,104 proteins detected, 70 were differentially expressed between baseline and the on-treatment time points; most of those 70 proteins are reflective of immune activation, Dr. Mehta said. They included PD-1, CXCL9, CXCL10, IL12Rb1, and others. Along with proteins reflective of an immune response, some involved regulation of cytokine secretion, negative regulation of T-cell proliferation, stress response, and other systems.
Thirty-eight of the proteins were differentially expressed between responders and nonresponders to immune checkpoint blockade. Some of the proteins that were enriched in nonresponders are specifically associated with resistance to immunotherapy, such as AREG and MIA. Other proteins that were differentially expressed between responders and nonresponders included CD1c, CXCL13, CLEC5a, IL8, and IGF1R.
Tissue biopsies were available from several of the patients in the cohort, from which immune cell infiltrates were processed for single-cell RNA sequencing. Using this tool, they showed that the proteins that were differentially expressed between responders and nonresponders were most highly expressed in myeloid cells within the tumor microenvironment. “A majority of these proteins are enriched within the dendritic cells and monocytes/macrophages,” Dr. Mehta said, adding that the dendritic cells were the dominant group in this analysis.
The next step in this research, Dr. Mehta said, will involve a validation cohort with increased power; this cohort includes 158 patients and 444 total samples. The investigators will attempt to build a machine learning predictor to make better use of the proteomic information; Dr. Mehta said the current cohort was too small to allow for such a tool. This second cohort will also involve building layered datasets with metabolite panels and exosome RNA-Seq; the researchers also hope to assess clonal hematopoietic mutations to further investigate the role of myeloid cells within the tumor microenvironment.
Jason J. Luke, MD, FACP, of the University of Chicago, was the discussant for the study. “There is a high unmet need for peripheral biomarkers,” he said. “The biomarkers that have the most weight at this point are tissue-based biomarkers. Obviously, we don’t have copious tumor tissue from every patient we see. Peripheral biomarkers for immunotherapy have been very difficult to come by so far.”
He said the finding that the proteins related to resistance were enriched in the myeloid cells is of particular interest. “What exactly are those cells, and could we start to integrate the contents of the peripheral blood into the likelihood of treatment response versus treatment resistance?” he asked. “It seems unlikely to me that it would be an absolute predictor, but in the context of those tissue-based biomarkers we already have, that could be of benefit.”
Dr. Luke highlighted several outstanding questions that remain from this study using proteomic analysis. For example, it will be useful to see how the peripheral blood proteomics correlate with that seen in intratumoral immune infiltrates. “What is the predictive value of these findings toward response relative to intratumoral biomarkers” such as IFNγ, PD-L1, TMB, and others, he asked. And importantly, “how might these findings be moved into the treatment prediction setting?”
– David Levitan
TMB Can Vary Temporally, Spatially Within Patients
Tumor mutational burden (TMB) can vary both within tumors and between tumors from different surgical events, according to a new study (Abstract 27). This intra- and intertumor variation occurs at a similar rate and may limit the use of TMB as a predictor of response to immune checkpoint inhibitors.
“TMB is an emerging biomarker,” said Tanner M. Johanns, MD, PhD, of the Washington University School of Medicine in St. Louis, who presented the study. (Dr. Johanns presented the results due to ASCO regulations that presentations not be given by researchers employed by industry; the study was led by Jeffrey M. Conroy, BS, chief scientific officer for OmniSeq.)
Dr. Johanns noted that some recent research has found intratumoral differences in TMB. “Does it matter which tissue we use for assessing TMB, from a time and distance standpoint?” he asked.
To assess that question, the researchers conducted a study including 36 patients with seven types of cancer; researchers collected 98 samples from these patients. Intratumor assessments, aimed to analyze whether spatial distribution of the tissue is relevant, included two to six samples from one surgical event (primary and/or metastatic sites). Intertumor assessments, which could determine if temporal differences in TMB are observed, involved one to three samples from one to three surgical events (primary or metastatic sites and sites of recurrence). There was a broad range in the time between surgical events, ranging from 5 months up to 96 months. TMB was measured using OmniSeq next-generation sequencing with a targeted 409-gene panel; TMB-high was considered any value above 10 muts/Mb.
The analysis revealed that TMB variation both within tumors and from different surgical events occurred in approximately 17% of patients. The specifics of that variation, or consistency, took many forms. For example, one patient had a skin metastasis with a TMB of 8.9 muts/Mb, and a lymph node sample taken 9 months later with a similar TMB of 8.0 muts/Mb. There was a 73% overlap in the specific variants seen in those samples.
Sometimes, similar TMBs were seen when the variant overlap was substantially lower. A sample from one patient’s brain lesion had a TMB of 3.6 muts/Mb compared with 4.6 muts/Mb in a small bowel metastatic lesion taken 19 months later; the variant overlap in this patient was only 29%. There were other, more extreme examples, including a patient with a primary skin lesion with a very high TMB of 51.6 muts/Mb and a metastatic lung lesion taken 34 months later with a TMB of 28.5 muts/Mb; there was a 0% variant overlap in these two samples.
Such discordance was also found in a few patients in the intratumor assessments. For example, one patient with colorectal cancer had a TMB in one sample of 38.7 muts/Mb, in another, a colon sample of 24.3 muts/Mb, and a much lower TMB of 3.6 muts/Mb in a rectal sample. There was a 3% variant overlap among these samples from one patient.
Other patients with samples from a single surgical event had very strong TMB correlation. One patient with colon cancer and liver metastases had four samples ranging from 9.7 muts/Mb to 10.7 muts/Mb and 73% variant overlap. Another, with ovarian cancer and adrenal gland metastases, had values ranging from 3.5 muts/Mb to 6.2 muts/Mb and variant overlap of 67%.
“Variation in overall TMB does seem to occur, but it is only in a minority of patients,” Dr. Johanns said, adding that the clinical interpretation of TMB will need to factor in issues like multisite sequencing and the influence of treatment.
Shailender Bhatia, MD, of the University of Washington Fred Hutchinson Cancer Research Center, was the discussant for the study. “TMB by itself is unlikely to be the biomarker that is going to solve our problems, but I think it is an important component … to further decipher which patients are going to respond,” he said.
Dr. Bhatia agreed with the conclusions that variation in overall TMB seems to occur in a minority of patients, although spatial and temporal variation are commonly seen.
“Is TMB ready for routine use in the clinic? I would argue not,” Dr. Bhatia said. “We need to continue our quest for an ideal predictive biomarker,” one that is reproducible, conserved across time and anatomic site, and that can reliably predict response or nonresponse.
– David Levitan