Search

The role of mass spectrometry-based serum proteomics signatures in predicting clinical outcomes in cancer patients treated with immune checkpoint inhibitors(ICI)Background Although immune checkpoint inhibitors (ICI) have changed the therapeutic scheme for multiple cancers, only a subset of patients experiences durable benefit. As current tumor markers such as PD-L1 show limited reliability in predicting clinical outcomes, we have explored the predictive role of markers representative of the host immune response on a systemic level present in the circulating proteome. Mass spectrometry allows for analysis of the proteome without the specific identification of individual proteins and protein isoforms. Here, we analyzed the most recent studies using mass spectrometry-based serum proteomics in predicting response to ICI treatment. Method A systematic literature search on Pubmed and major oncology scientific meetings was conducted up to November 18, 2020. Result Classifier algorithms trained with data sets from advanced stages of lung cancer (Primary Immune Response, Host Immune Classifier) and melanoma (BDX008, Immune Checkpoint Blockade) were used to stratify patients into groups with favorable and unfavorable treatment outcome. Patients with unfavorable predictive markers had worse prognosis when treated with ICI-single agent therapy. For patients treated with ICI alone or with chemotherapy as frontline or beyond, mass spectrometry-based serum proteomic signatures were shown to be a reliable predictive marker for survival outcomes (hazard ratios 0.15-0.5) independent of PD-L1 expression level. Conclusion Mass spectrometry-based serum proteomic tests reliably identify patients expected to have a worse prognosis. These patients can benefit from frontline aggressive treatment strategy combining ICI and chemotherapy rather than the standard of care ICI monotherapy. Cancer type Advanced stage NSCLCTreatmentLine of therapyClassifiernumber of patients included (n)number of patients in each classifier, n(%)Survival outcome (OS, month)HR [95%CI]referenceAdvanced stage NSCLCImmunotherapy (nivolumab)2nd linePIR116Not resistant75 (65%)17.30.48 [0.30-0.77], p=0.002Mirte Muller, et al.Resistant41(35%)6.0Sensitive32 (28%)11.10.58 [0.38-0.87], p=0.009Not sensitive84 (72%)4.3Immunotherapy ± ChemotherapyAll lines First line single agent immunotherapy (pembrolizumab) First line Combination (immunotherapy/chemotherapy) All lineHIC284Hot196 (69%)Not reached0.38 [0.27-0.53], p Citation Format: Yoonhee Choi, Jin Young Hwang, Won Kyung Hur, Myungwoo Nam, Leeseul Kim, Yeun Ho Lee, William Cheng, Eugene Kim, Emma Yu, Chan Mi Jung, William Han Bae, Young Kwang Chae. The role of mass spectrometry-based serum proteomics signatures in predicting clinical outcomes in cancer patients treated with immune check point inhibitors (ICI) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 673.

Background: Apart from its role in ovarian cancer progression, epithelial-mesenchymal transition (EMT) can promote chemotherapy resistance. We aimed to analyze how the EMT score would affect the tumor microenvironment in ovarian cancer. Methods: cBioportal was queried to obtain The Cancer Genome Atlas (TCGA) data for the serous ovarian cancer (SOC) cohort (TCGA, 585 patients). The neoantigen prediction data was derived from the CloudNeo pipeline using TCGA mutation calling. EMT scores were calculated by subtracting the average RNA-seq z-scores of three epithelial marker genes from the average RNA-seq z-scores of 13 mesenchymal marker genes as described in the previous study. Patient samples were grouped as either EMT-high (highest 1/3) or EMT-low (lowest 1/3). CIBERSORT was applied to predict the tumor-infiltrating immune cells. Results: Among the 585 SOC patients, only 256 patients had mutation data available for our analysis. The EMT-low group had a significantly higher mutation count (p value=0.0004) and cytolytic score (p value=0.032) than the EMT-high group. In addition, the EMT-low samples were associated with improved overall survival in SOC patients (HR, 0.55; 95% CI, 0.39-0.78; P < 0.001). The median survival of EMT-low was 57.40, and EMT-high was 41.06 months. Neoantigen counts and PD-L1 express level tended to be higher in the EMT-high group although failed to show statistical significance. The immune cell infiltration rates were not different between both groups. Conclusions: Our study is the first to describe the association between the EMT potential, neoantigen counts, and cytolytic scores in SOC. In our analyses, tumors with low EMT potential had a significantly higher neoantigen burden and higher cytolytic scores, suggesting that tumors with low EMT potentials tend to be more immunogenic. Further studies are warranted to explore the utility of EMT scores as biomarkers to predict the treatment response to immunotherapy in SOC. The immunologic characteristics of EMT low and high SOCNumber of patientsNeoantigen countp-valueMutation countp-valueCytolytic scorep-valuePD-L1 expressionp-valueEMT low6088.190.183783.190.0004183.40.032190.0663EMT high5867.7472.7486.3515.52 Citation Format: Won Kyung Hur, Jin Young Hwang, Leeseul Kim, Myungwoo Nam, William H. Bae, Yoonhee Choi, Yeun Ho Lee, Heayoon S. Cho, Emma Yu, Chan Mi Jung, William Cheng, Eugene Kim, Christmann Low, Victor Wang, Jeff Chuang, Young Kwang Chae. The neoantigen and immune landscape of epithelial mesenchymal transition (EMT) low and high score serous ovarian cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 626.

We aimed to understand how TGF-β signaling pathway related gene mutation affects the tumor neoantigen burden and immune landscape in uterine corpus endometrial carcinoma (UCEC). cBioPortal was queried to obtain the UCEC The Cancer Genome Atlas (TCGA) cohort data (521 patients). TGF-β signaling pathway related gene mutation was defined as genetic variances in TGFBR1, TGFBR2, ACVR2A, ACVR1B, SMAD2, SMAD3, or SMAD4. The neoantigen counts were predicted using CloudNeo pipeline. CIBERSORT was used to predict tumor-infiltrating immune cells. In 521 UCEC patients, 124 had mutations in TGF-β signaling pathway related genes. These tumors with the mutations had significantly higher neoantigen counts, mutation counts, and cytolytic scores (a geometric mean of mRNA expression of perforin and granzyme), and a trend towards higher PD-L1 expression level (Table 1). When subdivided by molecular subtypes, similar patterns were seen in most groups. TGF-β signaling pathway gene mutation was significantly associated with tumor tissue infiltration by CD8 T cells (18.3% vs 10.5%; p< 0.05), active CD4 memory T cells(13.0% vs 7.5%; p Immune landscape of endometrial carcinoma according to TGF-β pathway related gene mutationnumber of valuesneoantigenmutation countcytolytic scorePD-L1 expressionAll521166.1964.5189.914.1TGF-β pathway related gene nonmutated39652.3219.415813.4TGF-β pathway related gene mutated124528.4**3350.0**293.2*16.3POLEmutAll85741.04925268.619.0nonmutated28158.9941.6183.517.7mutated571027.0**6881.0**310.4*19.6MSI-HAll167297.51455.0251.814.8nonmutated99157.1539.5207.714.7mutated68501.9**2787.0**315.915.0P53abnAll192134.7904.1202.616.8nonmutated14820.390.8142.215.6mutated44519.3**3640.0**407.218.5NSMPAll16111.5254.9131.59.7nonmutated15411.554.6133.39.6mutated712.960.491.612.1Values are expressed as mean.*p Citation Format: William H. Bae, Jin Young Hwang, Won Kyung Hur, Myungwoo Nam, Yoonhee Choi, Leeseul Kim, Yeun Ho Lee, William Cheng, Eugene Kim, Emma Yu, Chan Mi Jung, Heayoon S. Cho, Jeff Chuang, Victor Wang, Young Kwang Chae. Immune landscape of endometrial carcinoma and its association with neoantigen landscape and TGF-β signaling pathway related genes mutation status [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 451.

Background: Within serous ovarian cancer (SOC), high-grade serous carcinoma (HGSC) demonstrates a worse prognosis compared to low-grade serous carcinoma (LGSC). Molecular analysis of SOC has revealed a distinct pattern of mutations found between the two pathologic subtypes, with TP53 as the main driver for HGSC. We analyzed the neoantigen and immune landscape of different SOC subtypes according to pathological grades and mutation status. Methods: The analysis was done using the SOC cohort of The Cancer Genome Atlas (TCGA). The neoantigen prediction was done through the CloudNeo pipeline. CIBERSORT was applied to derive the tumor-infiltrating immune cells. Patients were grouped by their pathological grades (high vs. low) and the molecular features (type 1 vs. 2). For further subgroup analysis, high-grade samples were stratified by their TP53 mutation status. Results: Among 585 SOC patients, a subset of 254 patients with available mutation counts and predicted neoantigen counts were included. There was no significant difference in neoantigen count between low and high grades (median 50.50 vs. 62.50, p=0.84). However, there was a trend for differential numbers in the neoantigen count between type 1 and type 2 (median 29 vs. 66.50, p=0.07). No notable differences were detected in immune landscapes of low vs. high grade and type 1 vs. type 2. Conclusions: Our study is the first to describe the neoantigen and immune landscape of SOC. There were no significant differences in the immune landscape between the subtypes. It seems that molecular subtyping is more related to neoantigen differences than histologic subtyping. It is likely that the neoantigen differences are not defined by the histologic grade but rather by molecular trait. The immunologic characteristics of EMT low and high SOCNumber of ptientsNeoantigen countp-valueMutation countp-valueCytolytic scorep-valuePD-L1 expressionp-valueLow grade1250.500.84700.8757.120.4812.340.28High grade8062.5085.571.0919.66Type 117290.07114.60.98125.10.7824.940.77Type 222066.50115.3119.526.3High grade(TP53 wild type)744.000.06620.08163.20.0629.770.5High grade(TP53 mutated)7364.0086110.126.47TCGA cohort provided the data of low(GB, G1, G2) and high(G3, G4) grade SOCsType 1: Defined as mutations in KRAS, BRAF, PTEN, PIK3CA, CTNNB1 and ARID1A irrespective of histologic gradeType 2: Defined as mutation in TP53 irrespective of histologic gradeThe samples that had intersecting mutations were excluded in type 1 and type 2.The cytolytic activity score was defined as a geometric mean of mRNA expression of perforin and granzyme. Citation Format: Won Kyung Hur, Jin Young Hwang, Leeseul Kim, Myungwoo Nam, William H. Bae, Yoonhee Choi, Yeun Ho Lee, William Cheng, Heayoon S. Cho, Emma Yu, Chan Mi Jung, Eugene Kim, Christmann Low, Victor Wang, Jeff Chuang, Young Kwang Chae. The neoantigen and immune landscape of low and high grade serous ovarian cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 453.

Early recognition of immune-related adverse events (irAEs) of immunotherapy is important. Circulating proteome reflects host response to diseases and is being explored as a marker for response to immunotherapy. We used a serum-based proteomics test, Primary Immune Response (PIR), to explore the associations between developing irAEs and immunotherapy in non-small cell lung cancer (NSCLC) patients. Data of 38 consented NSCLC patients with baseline PIR test done within one week prior to the start of immunotherapy were collected. Samples were grouped into either sensitive or intermediate/resistant (not sensitive) by PIR classification. We analyzed the durations from the immunotherapy initiation to the first episode of irAE, each individual irAE, and each irAE above grade 1 using log-rank test. IrAEs were graded according to Common Terminology Criteria for Adverse Events (CTCAE) v5.0. Among the 38 NSCLC patients, 21 patients (55%) experienced one or more irAEs. The total number of irAEs was 33 with the majority classified as either grade 1 (n=18, 55%) or grade 2 (n=11, 33%) (Table 1). PIR-sensitive group showed longer irAE free period with the median ‘Time to first irAE' being 54 weeks compared to 9.5 weeks in PIR-not sensitive (p=0.22, HR=0.56, 95% CI=0.24-1.34). The median ‘Time to each irAE' were 45 weeks and 12 weeks in PIR-sensitive and PIR-not sensitive, respectively (p=0.1, HR=0.55, 95% CI=0.28-1.1). The median ‘Time to each irAE above grade 1' demonstrated similar results with less differences between the two groups with median values of 54 weeks and 30 weeks in PIR-sensitive and PIR-not sensitive, respectively (p=0.28, HR=0.57, 95% CI=0.21-1.56). Our results demonstrated a trend that PIR-sensitive patients are more likely to tolerate immunotherapy longer without developing irAEs. It implies the potential value of the baseline PIR test in predicting the development of irAEs and selecting subsets of patients who need close monitoring with immunotherapy. Distribution of irAEs by PIR classificationVariablesSensitiveNot-sensitiveTotal number of patients, n1325Patients without irAE, n (%)7 (54%)10 (40%)Patients with irAE, n (%)6 (46%)15 (60%)Total number of irAEs, n1023Grade 1, n (%)5 (50%)13 (57%)Grade 2, n (%)5 (50%)6 (26%)Grade 3, n (%)02 (8%)Grade 4, n (%)01 (4%)Grade 5, n (%)01 (4%) Citation Format: Myungwoo Nam, Leeseul Kim, William Cheng, William H. Bae, Jin Young Hwang, Yoonhee Choi, Yeun Ho Lee, Won Kyung Hur, Chan Mi Jung, Heayoon S. Cho, Young Kwang Chae. Potential role of serum proteome in predicting immune-related adverse events from immunotherapy in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 520.