- •TMB alone is not sufficiently reliable or accurate as a biomarker of response to ICIs in NSCLC.
- •TMB-based survival prediction is improved by using the HLA-corrected TMB algorithm (TMB in combination with loss of heterozygosity of HLA).
- •Notably, additional predictive and prognostic value of the HLA-corrected TMB is not limited to certain types of cancer.
- •The HLA-corrected TMB could be a new strategy for selecting patients who may benefit from immunotherapy.
Immune checkpoint inhibitors (ICIs) have been shown to be beneficial for some patients with advanced non-small-cell lung cancer (NSCLC). However, the underlying mechanisms mediating the limited response to ICIs remain unclear.
Patients and methods
We carried out whole-exome sequencing on 198 advanced NSCLC tumors that had been sampled before anti-programmed cell death 1 (anti-PD-1)/programmed death-ligand 1 (PD-L1) therapy. Detailed clinical characteristics were collected on these patients. We designed a new method to estimate human leukocyte antigen (HLA)-corrected tumor mutation burden (TMB), a modification which considers the loss of heterozygosity of HLA from conventional TMB. We carried out external validation of our findings utilizing 89 NSCLC samples and 110 melanoma samples from two independent cohorts of immunotherapy-treated patients.
Homology-dependent recombination deficiency was identified in 37 patients (18.7%) and was associated with longer progression-free survival (PFS; P = 0.049). Using the HLA-corrected TMB, non-responders to ICIs were identified, despite having a high TMB (top 25%). Ten patients (21.3% of the high TMB group) were reclassified from the high TMB group into the low TMB group. The objective response rate (ORR), PFS, and overall survival (OS) were all lower in these patients compared with those of the high TMB group (ORR: 20% versus 59%, P = 0.0363; PFS: hazard ratio = 2.91, P = 0.007; OS: hazard ratio = 3.43, P = 0.004). Multivariate analyses showed that high HLA-corrected TMB was associated with a significant survival advantage (hazard ratio = 0.44, P = 0.015), whereas high conventional TMB was not associated with a survival advantage (hazard ratio = 0.63, P = 0.118). Applying this approach to the independent cohorts of 89 NSCLC patients and 110 melanoma patients, TMB-based survival prediction was significantly improved.
HLA-corrected TMB can reconcile the observed disparity in relationships between TMB and ICI responses, and is of predictive and prognostic value for ICI therapies.
Antibodies targeting the programmed cell death 1 (PD-1)/programmed death-ligand 1 (PD-L1) and cytotoxic T lymphocyte antigen 4 (CTLA-4) axes have greatly increased the survival of patients with a wide variety of cancer types.
4However, despite the transformative potential of immune checkpoint inhibitors (ICIs) in clinical care, only a few patients derive clinical benefit from this treatment, thereby highlighting the need for an effective biomarker for predicting the response to ICIs.
Several studies have consistently revealed an association between a high tumor mutation burden (TMB) and increased clinical efficacy following ICIs in a variety of cancer types.
8However, the clinical utility of TMB for identifying patients who may benefit from ICIs remains unclear. The distribution of TMB has been shown to substantially overlap between responsive and non-responsive tumors, indicating that TMB alone is not a sufficient biomarker for accurately predicting clinical responses.
Recent genomic studies have explored other potential biomarkers and evidence has emerged that alteration in the antigen presentation pathway may result in immune evasion and may affect antitumor immunity.
13However, the clinical implications of disrupted antigen presentation in response to ICIs has not been well-characterized.
12In this study, we analyzed whole-exome sequencing data from 198 advanced non-small-cell lung cancer (NSCLC) tumors sampled before immunotherapy. We investigated the clinical utility of genomic features related to the anticancer immune response as a predictive biomarker. We also applied our newly designed computational methods to characterize the associations between an altered tumor antigen presentation and clinical response to ICIs.
Patients and study assessments
Patients with advanced NSCLC who were administered anti-PD-1/PD-L1 therapy were eligible for enrollment in this study if they had valid data available for efficacy analysis and available biopsy tissue before the initiation of ICIs. A total of 198 patients with advanced NSCLC were included in this study from July 2014 to August 2018 (supplementary Figure S1A and Table S1, available at Annals of Oncology online). The study was approved by the Institutional Review Board of each participating hospital and all participating patients provided written informed consent before enrollment. To validate the findings derived from this study cohort, we obtained and analyzed exome sequencing data from two independent validation cohorts in which 89 NSCLC patients were treated with ICIs
13and 110 melanoma patients were treated with CTLA-4 blockade.
9Additional details regarding these patient cohorts are presented in the supplementary material of previously published works.
The objective response (OR) was assessed by using Response Evaluation Criteria in Solid Tumors, version 1.1 (RECIST v1.1). We considered patients showing complete response (CR) or partial response (PR) as responders, whereas patients with stable disease (SD) or progressive disease (PD) were considered non-responders. In addition to OR, we also categorized patient data into durable clinical benefit (DCB; CR/PR or SD lasting longer than 6 months) and no durable benefit (NDB; all other patients).
Whole-exome sequencing and exome analysis pipeline
All tumor samples (formalin-fixed paraffin-embedded and fresh frozen tissues) were obtained before ICIs treatment. DNA extraction, library preparation, and sequencing were conducted as described previously.
15The presence of tumor tissue and percentage of viable tumor in the sequenced samples were reviewed by a thoracic pathologist (YLC). The mean sequencing coverages across all tumor samples and blood samples were X158 and X105, respectively. Detailed methods for all components of the genetic analysis are provided in the supplementary Methods (available at Annals of Oncology online).
PD-L1 expression in the samples was assessed based on the US Food and Drug Administration-approved Dako PD-L1 IHC 22C3 pharmDx kit (Agilent Technologies, Santa Clara, CA).
Human leukocyte antigen-corrected tumor mutational burden algorithm
Currently available algorithms for in silico neoantigen prediction are based on predicted peptide-human leukocyte antigen (HLA) binding affinity.
18The performance of these methods can be improved by accumulating eluted ligand data; however, the neoantigen burden determined by computational prediction shows variable accuracy and is inadequate for predicting the response to ICIs.
17TMB has been shown to be more predictive of response to ICIs when compared with computationally derived neoantigen load.
Thus, rather than using the neoantigen burden, we determined an HLA-corrected TMB using the following equation:
TMB: number of nonsynonymous alteration (single-nucleotide variations or indels), NeoAg: computationally predicted neoantigen burden, as output by Mupexi,
16NeoAgL: neoantigens predicted to bind to the lost HLA alleles, as output by Mupexi and LOHHLA.
10The value was set to zero for cases without loss of heterozygosity of HLA (HLA-LOH). NeoAgC: neoantigens predicted to bind to both lost HLA alleles and the kept HLA alleles. The value was set to zero for cases without HLA-LOH.
The Kaplan–Meier method was used to estimate progression-free survival (PFS) and overall survival (OS). The log-rank test was used to assess differences between groups in PFS and OS. Hazard ratios and 95% confidence intervals (CIs) were computed using the Cox proportional hazards model. Categorical variables between two groups were compared using the Fisher's exact test or chi-square test for three groups. Differences in means or medians for a continuous variable between two groups were assessed by the non-parametric Mann–Whitney U test or unpaired t-test. All statistical analyses were carried out using R3.6.0 software (R Foundation for Statistical Computing, Vienna, Austria) and outlines in detailed in supplementary Methods (available at Annals of Oncology online).
Clinical and genomic characteristics of the study cohorts
Most patients were administered PD-1 inhibitor monotherapy (152/198, 77%), whereas the remainder were administered anti-PD-L1 monotherapy (46/198, 23%). The baseline clinical characteristics of the 198 patients are summarized in Table 1.
Table 1Baseline clinical characteristics
|N = 198||All patients||CR/PR||SD/PD||P value|
|No. (%)||61 (31%) No. (%)||137 (69%) No. (%)|
|Median age (range)||62.1 (33–84)||65.1 (44–83)||61.4 (33–84)||0.024|
|Male||140 (71)||48 (75)||94 (69)|
|Female||58 (29)||15 (25)||43 (31)|
|Median TMB (range)||143 (1–1765)||194 (3–1765)||131 (1–1035)||0.007|
|Current/former||130 (66)||45 (74)||85 (62)|
|Never||68 (34)||16 (26)||52 (38)|
|ECOG 0 & 1||172 (87)||56 (92)||116 (85)|
|ECOG 2||26 (13)||5 (8)||21 (15)|
|LUAD||129 (65)||40 (66)||89 (65)|
|LUSC||58 (29)||18 (29)||40 (29)|
|Others||11 (6)||3 (5)||8 (6)|
|Nivolumab||74 (37)||20 (33)||54 (40)|
|Pembrolizumab||78 (40)||27 (44)||51 (37)|
|Anti-PD-L1 agent||46 (23)||14 (23)||32 (23)|
|Line of therapy||0.393|
|First||14 (7)||5 (8)||9 (6)|
|Second||66 (33)||24 (39)||42 (31)|
|Third or more||118 (60)||32 (53)||86 (63)|
|<1%||34 (17)||8 (13)||26 (19)|
|1–49%||35 (18)||6 (10)||29 (21)|
|≥50%||75 (38)||31 (51)||44 (32)|
|Unknown||54 (27)||16 (26)||38 (28)|
| EGFR mut|
| KRAS mut|
|BRAF mut (p.V600E)||4||1||3|
CR, complete response; ECOG, The Eastern Cooperative Oncology Group; HLA LOH, loss of heterozygosity at the class I human leukocyte antigen; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; PD, progressive disease; PD-1, programmed cell death 1; PD-L1, programmed death-ligand 1; PR, partial response; SD, stable disease; TMB, tumor mutation burden.
a EGFR mutation: Del19, L858R, Del18, and Ins20.
b KRAS mutation: G12A, G12C, G12D, G12V.
The objective response rate (ORR = CR/PR) to single-agent ICI was 31% (61/198); DCB rate was 36% (72/198). The median TMB was 143 mutations (range, 1–1765 mutations). The distribution of TMB was similar to that in patients with NSCLC treated with ICIs in a previous study (n = 75; supplementary Figures S1B and S1C, available at Annals of Oncology online).
2Of the 198 patients, 184 patients (93%) were administered subsequent systemic anticancer therapy. The Eastern Cooperative Oncology Group (ECOG) performance status score was similar between OR and no response (SD/PD), but 26 patients (13%) with an ECOG score of 2 exhibited significantly shorter PFS (hazard ratio = 2.43, 95% CI, 1.54–3.84; log-rank test, P < 0.0001; Figure 1A; supplementary Figure S2A, available at Annals of Oncology online) and OS (hazard ratio = 4.95, 95% CI, 2.98–8.21; log-rank test, P < 0.0001; supplementary Figure S2B, available at Annals of Oncology online). Other clinical characteristics were evenly distributed between the OR group and no response group (supplementary Figure S2C–F, available at Annals of Oncology online).
Association between TMB and response to ICIs
Consistent with prior reports, TMB was significantly higher in tumors of responders compared with that of the no response group (median TMB 194 versus 131 mutations, Mann–Whitney test, P = 0.0069; Figure 1B and supplementary Figure S1E, available at Annals of Oncology online).The variable cut-off of TMB was used to define ‘high TMB’ in previous studies.
6We defined TMB subgroups by the 25 percentile value of the cohort combined with the independent NSCLC cohort treated with ICIs (Hellmann cohort; supplementary Figure S1C, available at Annals of Oncology online).
2We used this approach to obtain a more universal cut-off point for high TMB. There was a significant association between high TMB (top 25% of the two independent cohorts, 272 mutations) and improved PFS (hazard ratio = 0.67, 95% CI, 0.45–0.99; log-rank test, P = 0.043; Figure 1D). However, the area under the curve (AUC) for correlation of TMB with OR was 0.62 (95% CI, 0.53–0.71; supplementary Figure S1D, available at Annals of Oncology online) and with DCB was 0.56 (95% CI, 0.48–0.65).
HR gene alteration is associated with elevated TMB and clinical benefit from ICIs
HR deficiency was observed in 37 patients (18.7%) and was associated with longer PFS in Kaplan–Meier survival analysis (hazard ratio = 0.65, 95% CI, 0.42–1.00; log-rank test, P = 0.049; Figure 1C and E; supplementary Figures S2H and S3C, available at Annals of Oncology online). To clarify the association of HR deficiency with ICI response, we investigated whether this association remained significant in a multivariable model. PD-L1 expression data were available for 144 patients. Altered HR function (27 of 144, 18.8%) remained as an independent predictor of the response to ICIs after adjusting for age, sex, ECOG score, histology, smoking, and PD-L1 expression in these patients (hazard ratio = 0.58, 95% CI, 0.34–0.99, P = 0.046; supplementary Figure S3A, available at Annals of Oncology online).
PD-L1 expression and response to ICIs
Patients with high (≥50%) PD-L1 expression showed an ORR of 42% (31/75) and DCB rate of 44% (33/75). PD-L1 expression was significantly higher among patients showing a favorable OR compared with non-responders (CR/PR versus SD/PD, Mann–Whitney test, P = 0.0020). Similar results were seen in patients with DCB versus NDB (Mann–Whitney test, P = 0.0266). Like TMB, PD-L1 expression alone was not sufficient to predict a response (PD-L1 AUC = 0.66, 95% CI, 0.56–0.76; supplementary Figure S1F, available at Annals of Oncology online) or clinical benefit (PD-L1 AUC = 0.61, 95% CI, 0.51–0.71).
Mutations in antigen presentation machinery and response to ICIs
We investigated whether somatic mutations and copy number alterations contribute to the restricted response to ICIs (supplementary Figure S4, available at Annals of Oncology online). STK11 mutation was observed in nine patients; however, the association with the response to ICIs was not significant (supplementary Figure S3B, available at Annals of Oncology online). We also investigated mutations in components of the antigen presentation machinery. Unfortunately, truncating or deleterious mutations in antigen presentation machinery were too rare to assess their associations (supplementary Figure S4C, available at Annals of Oncology online). Similarly, we did not identify a significant association between focal copy number alterations and clinical benefit.
LOH at class I HLA locus and TMB
We observed 54/198 (27%) tumors with LOH in at least one HLA-I locus (Figure 2A) and HLA-LOH was associated with a significant increase in somatic nonsynonymous mutations consistent with prior studies (median TMB 199 versus 118 mutations, Mann–Whitney test, P = 0.0008; Figure 2C).
19Consistent with prior reports, there was no association between HLA-LOH and response to the anti-PD-(L)1 agent (Figure 2B and F).
We investigated whether the elevated mutational load among patients with HLA-LOH was attributable to neoantigen binding to the lost HLA alleles. We designed a computationally HLA-corrected TMB based on the proportion of neoantigens predicted to bind to the kept HLA alleles (see Methods; Figure 2D). Our results showed that computationally HLA-corrected TMB was not significantly elevated in patients with HLA-LOH compared with those without HLA-LOH (median HLA-corrected TMB 149 versus 118 mutations, Mann–Whitney test, P = 0.2061; Figure 2E).
HLA-corrected TMB with HLA-LOH has additional predictive benefits for response to ICIs
We applied the HLA-corrected TMB algorithm to samples with HLA-LOH and observed that 10 patients (21.3% of the high TMB group) were reclassified from the high TMB group (top 25%, 272 mutations) into the low TMB group (<25%; Figure 3A and B). Interestingly, ORR, DCB rate, PFS, and OS were all lower in patients showing changes in the TMB subgroup (subgroup-reclassified patients) compared with those in patients of the high TMB group (ORR 20% versus 59%, Fisher's exact test, P = 0.0363; DCB 20% versus 59%, Fisher's exact test, P = 0.0363; PFS hazard ratio = 2.91, 95% CI, 1.29–6.54; log-rank test, P = 0.007; OS hazard ratio = 3.43, 95% CI, 1.41–8.34; log-rank test, P = 0.004; Figure 3C–F). As an exploratory analysis, we compared the subgroup-reclassified patients (n = 10; TMB range 282–477 mutations; HLA-corrected TMB 143–271 mutations) with the top 15%–25% of the TMB group without HLA-LOH (n = 9; median TMB 308 mutations). Although TMB in the subgroup-reclassified patients was similar to that in the top 15–25% of the TMB group (median TMB 320 versus 308 mutations, Mann–Whitney test, P = 0.5953; supplementary Figure S5A and S5B, left, available at Annals of Oncology online), PFS and OS were all lower in the subgroup-reclassified patients (PFS hazard ratio = 3.40, 95% CI, 1.02–11.42; log-rank test, P = 0.037; OS hazard ratio = 7.12, 95% CI, 1.44–35.28, log-rank test, P = 0.007; supplementary Figure S5C and S5D, available at Annals of Oncology online); the HLA-corrected TMB for the subgroup-reclassified patients was significantly different from that of the top 15%–25% group (median HLA-corrected TMB 221 versus 308 mutations, Mann–Whitney test, P < 0.0001; supplementary Figure S5B, right, available at Annals of Oncology online).
We next assessed whether the HLA-corrected TMB has further predictive advantage beyond the conventional TMB in multivariable models. Our results showed that HLA-corrected TMB was predictive of improved PFS and OS in the multivariable analysis after controlling for age, sex, smoking status, ECOG score, and PD-L1 expression (HLA-corrected TMB high, PFS hazard ratio 0.45; 95% CI, 0.26–0.78; P = 0.004; OS hazard ratio = 0.44, 95% CI, 0.23–0.85; P = 0.015; Table 2), whereas conventional TMB was not associated with a survival advantage (conventional TMB high: PFS hazard ratio 0.58; 95% CI, 0.35–0.94; P = 0.028; OS hazard ratio = 0.63, 95% CI, 0.36–1.12; P = 0.118; supplementary Table S2 and Figure S2G, available at Annals of Oncology online).
Table 2Multivariate Cox regression analyses of overall survival among patients with available for PD-L1 expression
|Stratified by conventional TMB||Stratified by HLA-corrected TMB|
|Overall survival (N = 144)|
|Overall survival (N = 144)|
|Covariate||Multivariate analysis||Covariate||Multivariate analysis|
|HR (95% CI)||P||HR (95% CI)||P|
|Age||<65 yr||Reference||—||Age||<65 yr||Reference||—|
|≥65 yr||1.20 (0.72–1.98)||0.483||≥65 yr||1.18 (0.72–1.94)||0.518|
|Male||1.51 (0.58–3.98)||0.402||Male||1.47 (0.57–3.84)||0.427|
|Performance status||ECOG 0 & 1||Reference||—||Performance status||ECOG 0 & 1||Reference||—|
|ECOG 2||6.50 (3.49–12.13)||<0.0001||ECOG 2||6.45 (3.49–11.93)||<0.0001|
|Smoking status||Never||Reference||—||Smoking status||Never||Reference||—|
|Current/former||1.44 (0.59–3.50)||0.420||Current/former||1.57 (0.65–3.78)||0.319|
|TMB group||TMB low||Reference||—||HLA-corrected TMB group||HLA-corrected TMB Low||Reference||—|
|TMB high||0.63 (0.36–1.12)||0.118||HLA-corrected TMB high||0.44 (0.23–0.85)||0.015|
|PD-L1 expression||<1%||Reference||—||PD-L1 expression||<1%||Reference||—|
|1–49%||1.50 (0.78–2.87)||0.227||1–49%||1.57 (0.81–3.01)||0.179|
|≥50%||1.23 (0.67–2.29)||0.505||≥50%||1.21 (0.66–2.23)||0.543|
CI, confidence interval; ECOG, The Eastern Cooperative Oncology Group; HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1; TMB, tumor mutation burden.
a Result depicts only those patients with available for PD-L1 expression (n = 144).
To validate this finding, we further analyzed 89 NSCLC samples and 110 melanoma samples from two independent ICI-treated cohorts.
13Our reanalysis of the 89 NSCLC tumor samples treated with ICIs revealed that, consistent with the findings of our current study, TMB-based survival prediction was improved by using the HLA-corrected TMB algorithm (log-rank test P = 0.048 for conventional TMB and log-rank test P = 0.008 for HLA-corrected TMB; Figure 4 and supplementary Figure S6, available at Annals of Oncology online). Similar to the NSCLC patients cohort, the association between HLA-corrected TMB and improved survival prediction was also observed when we applied the HLA-corrected TMB algorithm to melanoma patients in multivariable models (HLA-corrected TMB high: OS hazard ratio 0.54; 95% CI, 0.31–0.94; P = 0.030; conventional TMB high: OS hazard ratio 0.60; 95% CI, 0.35–1.02; P = 0.060; supplementary Figure S7 and Table S3, available at Annals of Oncology online).
We carried out a comprehensive genomic analysis of advanced NSCLC samples to investigate the role of tumor genomic and antigen presentation features in determining the response to ICIs. We observed that genetic features associated with the anticancer immune response, including HLA-LOH and HR deficiency, can impact the response to ICIs. Our study cohort showed that HLA-LOH alone is not associated with an inferior clinical response to antibodies targeting PD-1/PD-L1. This result is consistent with the findings of Rodig et al.,
12which indicated that major histocompatibility complex (MHC) class I expression was not associated with primary resistance to anti-PD-1 agents among patients with melanoma. Our reanalysis of the 110 melanoma samples (Van Allen cohort) also support the observation that presence of HLA-LOH is not associated with an inferior clinical response (supplementary Figure 7C, available at Annals of Oncology online).
Because of the unexpected discrepancy between the observed results and immune surveillance hypothesis, we investigated whether the association of HLA-LOH and response to ICIs would be better characterized by considering TMB. Interestingly, tumors with HLA-LOH showed a higher mutational burden compared with tumors without HLA-LOH, whereas the HLA-corrected TMB was not significantly elevated in tumors with HLA-LOH. This observation supports the hypothesis that HLA-LOH allows for subsequent subclonal expansion.
10Previous studies identified elevated TMB in tumor samples exhibiting HLA-LOH and confirmed that an elevated TMB is attributed to subclonal neoantigens that are predicted to bind to the lost HLA alleles.
We further found that TMB-based response prediction was improved by using the HLA-corrected TMB algorithm. The observed outlier in the relationship between a high TMB and ICI response was reconciled by considering the HLA-LOH. It was hypothesized that neoantigens predicted to bind only to lost HLA alleles may not elicit effective antitumor immunity. This result is consistent with previous reports that show subclonal mutation may not drive effective tumor immune responses.
20The HLA-corrected TMB can help to identify patients who do not respond to ICIs despite having a high TMB. Based on our observations of 89 NSCLC patients and 110 melanoma patients from independent ICI-treated cohorts, our findings suggest that the clinical utility of HLA-corrected TMB could be broadly applied to various cancer types. Although further data are needed to confirm this observation, HLA-corrected TMB with HLA-LOH have additional predictive and prognostic value for response to ICIs.
Our results also showed that HR gene alteration was associated with a higher TMB and longer PFS. Previous analyses suggest that cytosolic DNA fragments derived from defective DNA damage response and repair mechanisms influence the response to ICIs by triggering the stimulator of interferon genes (STING) signaling pathway, which highlights the additional influence of DNA repair alterations on the checkpoint blockade response.
22Various cytoplasmic DNA sensors may bind cytosolic DNA fragments and activate the STING pathway, which induces antitumor activity via type 1 interferon and T cell recruitment.
23Notably, Kitajima et al.
24found that STK11 loss leads to immune evasion through methylation-induced suppression of STING. Indeed, STK11-mutation was associated with primary resistance to ICIs in KRAS-mutant lung adenocarcinoma cohorts.
One limitation of this study is that numerous patients were administered subsequent systemic anticancer therapies. Additionally, patients with an ECOG score of 2 were included, causing the median PFS and OS in this cohort to be shorter than those reported previously.
8However, given that most patients are administered systemic anticancer therapies in real clinical settings, data from this cohort may reflect the real-world use of ICIs in patients with advanced NSCLC.
In conclusion, through integrated analysis of genomic and clinical data, we evaluated the observed disparity in the relationship between genomic biomarkers and the response to ICIs. We suggest that the HLA-corrected TMB, which takes HLA-LOH into account, could be a new strategy for selecting patients who may benefit from immunotherapy. Although further studies are needed, our findings provide insight into the underlying molecular features contributing to response to ICIs and refine our understanding of the antitumor immune mechanisms of ICIs.
We would like to acknowledge all of the patients and their families for their contributions to this study. We thank Matthew L. Meyerson, Sam Freeman, and Tao Zou at the Dana-Farber Cancer Institute for their critical review of the manuscript. We are grateful to Sang Ha Shin and Tae Hee Hong at Samsung Genome Institute who provided engaging discussions and critical review of our manuscript.
This work was supported by the Post-Genome Technology Development Program (Business model development driven by the clinico-genomic database for precision immuno-oncology) funded by the Ministry of Trade, Industry and Energy (MOTIE, Korea) [grant number 10067758 ]; and a grant from the National Research Foundation of Korea funded by the Korean Government (NRF-2018-Global PhD Fellowship Program) [grant number 2018H1A2A1062330 ] to JHS. This work was supported in part by US National Institutes of Health grants [grant number CA121113 ] (VA), the Eastern Cooperative Oncology Group-American College of Radiology Imaging Network (VA), the V Foundation (VA), Swim Across America (VA), the Allegheny Health Network — Johns Hopkins Research Fund (VA), and the LUNGevity Foundation (VA) (no grant numbers).
JSA reports personal fees from Amgen, personal fees from Pfizer, personal fees from AstraZeneca, personal fees from Menarini, personal fees from Roche, personal fees from Eisai, personal fees from Boehringer Ingelheim, personal fees from Bristol-Myers Squibb-Ono, personal fees from Merck Sharp & Dohme (MSD), personal fees from Janssen, personal fees from Samsung Bioepis, outside the submitted work. S-HL reports grants and personal fees from MSD, personal fees from Novartis, personal fees from AstraZeneca, personal fees from Bristol-Myers Squibb, personal fees from Roche, outside the submitted work. KP reports personal fees from Astellas, Astra Zeneca, AMGEN, Boehringer Ingelheim, Clovis, Eli Lilly, Hanmi, KHK, Merck, MSD, Novartis, ONO, Roche, BluePrint, outside the submitted work. VA receives research funding from Bristol-Myers Squibb. All remaining authors have declared no conflicts of interest.
- Supplementary Methods
- Supplementary Table S1
- Supplementary Table S2
- Supplementary Table S3
- Supplementary Figure Legends
- Supplementary Figure S1
- Supplementary Figure S2
- Supplementary Figure S3
- Supplementary Figure S4
- Supplementary Figure S5
- Supplementary Figure S6
- Supplementary Figure S7
- The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy.Nat Rev Cancer. 2019; 19: 133-150
- Genomic features of response to combination immunotherapy in patients with advanced non-small-cell lung cancer.Cancer Cell. 2018; 33: 843-852.e4
- Tumor mutational load predicts survival after immunotherapy across multiple cancer types.Nat Genet. 2019; 51: 202-206
- Genomic correlates of response to immune checkpoint blockade.Nat Med. 2019; 25: 389-402
- STK11/LKB1 mutations and PD-1 inhibitor resistance in KRAS-mutant lung adenocarcinoma.Cancer Discov. 2018; 8: 822-835
- Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic.Ann Oncol. 2019; 30: 44-56
- Genomic correlates of response to immune checkpoint blockade in microsatellite-stable solid tumors.Nat Genet. 2018; 50: 1271-1281
- Molecular determinants of response to anti-programmed cell death (PD)-1 and anti-programmed death-ligand 1 (PD-L1) blockade in patients with non-small-cell lung cancer profiled with targeted next-generation sequencing.J Clin Oncol. 2018; 36: 633-641
- Genomic correlates of response to CTLA-4 blockade in metastatic melanoma.Science. 2015; 350: 207-211
- Allele-specific HLA loss and immune escape in lung cancer evolution.Cell. 2017; 171: 1259-1271.e11
- Patient HLA class I genotype influences cancer response to checkpoint blockade immunotherapy.Science. 2018; 359: 582-587
- MHC proteins confer differential sensitivity to CTLA-4 and PD-1 blockade in untreated metastatic melanoma.Sci Transl Med. 2018; 10: eaar3342
- Multimodal genomic features predict outcome of immune checkpoint blockade in non-small-cell lung cancer.Nat Cancer. 2020; 1: 99-111
- Comprehensive molecular characterization of clinical responses to PD-1 inhibition in metastatic gastric cancer.Nat Med. 2018; 24: 1449-1458
- Prevalence and detection of low-allele-fraction variants in clinical cancer samples.Nat Commun. 2017; 8: 1377
- MuPeXI: prediction of neo-epitopes from tumor sequencing data.Cancer Immunol Immunother. 2017; 66: 1123-1130
- NetMHCpan-4.0: improved peptide–MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data.J Immunol. 2017; 199: 3360-3368
- HLA class I alleles are associated with peptide-binding repertoires of different size, affinity, and immunogenicity.J Immunol. 2013; 191: 5831-5839
- Allele-specific HLA loss and immune escape in lung cancer evolution.Cell. 2017; 171: 1259-1271.e11
- Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade.Science. 2016; 351: 1463-1469
- Alterations in DNA damage response and repair genes as potential marker of clinical benefit from PD-1/PD-L1 blockade in advanced urothelial cancers.J Clin Oncol. 2018; 36: 1685-1694
- DNA damage and innate immunity: links and trade-offs.Trends Immunol. 2014; 35: 429-435
- STING recognition of cytoplasmic DNA instigates cellular defense.Mol Cell. 2013; 50: 5-15
- Suppression of STING associated with LKB1 loss in KRAS-driven lung cancer.Cancer Discov. 2019; 9: 34-45
Published online: April 19, 2020
© 2020 The Author(s). Published by Elsevier Ltd on behalf of European Society for Medical Oncology.
User LicenseCreative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) |
How you can reuse
Elsevier's open access license policy
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)
For non-commercial purposes:
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article (private use only, not for distribution)
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
- Distribute translations or adaptations of the article
Elsevier's open access license policy