Sage Journals: Discover world-class research

Abstract

Background:

Tumor mutation burden (TMB) has been validated as a predictive biomarker for immunotherapy response and survival in numerous cancer types. Limited data is available on the inherent prognostic role of TMB in early-stage tumors.

Objective:

To evaluate the prognostic role of TMB in early-stage, resected non-small cell lung cancer (NSCLC).

Design:

Systematic review and meta-analysis of pertinent prospective and retrospective studies.

Data sources and methods:

Publication search was performed in PubMed, Embase, Cochrane Library, and Web of Science databases. Based on the level of heterogeneity, a random- or fixed-effects model was used to calculate pooled effects of hazard ratio (HR) for overall survival (OS) and disease-free survival (DFS). The source of heterogeneity was investigated using sensitivity analysis, subgroup analysis, and publication bias assessment.

Results:

Ten studies comprising 2520 patients were included in this analysis. There was no statistically significant difference in OS (HR, 1.18, 95% CI, 0.70, 1.33; p 0.53, I² = 80%; p_het < 0.0001) and DFS (HR, 1.18, 95% CI, 0.91, 1.52; p = 0.53, I² = 75%; p_het = 0.0001) between the high-TMB and low-TMB group. Subgroup analyses indicated that East Asian ethnicity, and TMB detected using whole exome sequencing, and studies with <100 patients had poor DFS in the high-TMB group.

Conclusion:

The inherent prognostic role of TMB is limited in early-stage NSCLC. Ethnic differences in mutation burden must be considered while designing future trials on neoadjuvant immunotherapy. Further research in the harmonization and standardization of panel-based TMB is essential for its widespread clinical utility.

Registration: CRD42023392846.

Keywords

disease-free survival non-small cell lung cancer meta-analysis overall survival tumor mutation burden

Introduction

Lung cancer (LC) comprises molecularly and histologically distinct subtypes, of which non-small cell lung cancer (NSCLC) accounts for 85% of new lung cancer cases.¹ Approximately 39% of new NSCLC cases are diagnosed at an early stage (I-IIIA).² Despite curative treatment, the 5-year overall survival (OS) rate plunges from 92% in clinical stage IA1 to 36% in stage IIIA.³ The Tumor Nodes Metastasis (TNM) staging system is the standard prognostic assessment tool currently used in the clinical practice for LC.⁴ Due to tumor biological heterogeneity, anatomical classification inherently has deficiencies; therefore, reliable biomarkers for prognosis must be continuously sought.⁵

The number of somatic non-synonymous mutations within a tumor genome represents tumor mutational burden (TMB).⁶ Lung adenocarcinomas (LUAD) and squamous cell carcinomas (LSCC) have one of the highest prevalence of somatic mutations [above 10 somatic mutations per megabase (Mb) (Mut/Mb) of coding DNA], and thus, more likely to form neoantigens to be detected by autologous T-cells.⁷ Based on this phenomenon, TMB has been explored extensively as a biomarker of immunotherapy response in various cancer types.^8–10 High-TMB patients showed a significantly improved OS, progression-free survival (PFS), and objective response rate (ORR) following immune checkpoint inhibitor (ICI) therapy in advanced NSCLC, independent from programed cell death ligand-1 (PD-L1) expression.^9,11–13 Interestingly, a recent meta-analysis (10 studies, 491 participants) identified TMB as a promising predictive biomarker for major pathological response [Odds Ratio (OR), 3.40, 95% confidence interval 1.33–8.70; p = 0.0109] and pathological complete response (pCR) [OR, 1.98, 95% CI 1.08–3.63, p = 0.0265] to neoadjuvant immunotherapy in early-stage NSCLC, suggesting an emerging role of TMB in early-stage NSCLC.¹⁴

The inherent prognostic implication of TMB, independent of ICIs, in early-stage, resected NSCLC is variable, where some studies have shown a positive¹⁵ or negative association^16,17 between high-TMB and survival, while others have not.^18,19 Ethnicity, NSCLC subtypes, TMB detection methods, including the different regions of genes analyzed, cutoff values, and sample sources may further influence this association in individual studies. Consequently, we performed a systematic review and meta-analysis of studies that assessed the prognostic role of TMB for early-stage resected NSCLC survival.

Materials and methods

The Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) guidelines were utilized to conduct this systematic review and meta-analysis²⁰ (Supplemental Table S1). An a priori-defined protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42023392846).

Literature search

PubMed, Embase, Web of Science, and Cochrane Library were comprehensively searched for articles through 31 January 2023. The following Medical Subject Headings (MeSH)/Emtree terms were used (tumor mutational burden OR tumor mutation burden OR tumor mutational burden OR tumor mutational load OR tumor mutation load OR TMB OR TML) AND (lung cancer OR lung neoplasms OR lung tumor OR non-small cell lung cancer OR cancer of lung OR lung adenocarcinoma OR adenocarcinoma of lung OR squamous cell carcinoma of lung OR lung squamous cell carcinoma OR human pulmonary neoplasm OR human pulmonary carcinomas; Supplemental Table S2). To include a maximum number of appropriate studies, the American Society of Clinical Oncology, European Society for Medical Oncology (ESMO), and World Lung Cancer Congress conference abstracts were also searched. In addition, the references of the full-text eligible studies were consulted to collect as many suitable studies as possible.

Study selection

Article titles, abstracts, and full text (if required) were screened by two independent reviewers (DW and PH) and evaluated based on the inclusion criteria. In case of differing opinions, disagreements were resolved by mutual discussion or involving the third reviewer (SG).

The inclusion criteria were as follows: (1) Studies with patients diagnosed with histologically confirmed early-stage NSCLC (I-IIIA), (2) Studies reporting TMB into two or three groups (high TMB and/or intermediate TMB versus low TMB levels) based on clear TMB cut-off values, (3) Studies with OS and/or DFS as the clinical endpoints (4) Studies with either reported or extractable time-to-event data as HR and its 95% CI related with the TMB level,²¹ (5) No language restriction was placed.

The PICOS criteria for this review were as follows: Participants: adults with histologically confirmed primary early-stage NSCLC; Intervention: curative intent surgical resection (lobectomy, pneumonectomy, segmentectomy, and sublobar resection) along with mediastinal lymph node evaluation (sampling, systematic lymph nodal dissection, and lobe-specific lymph nodal dissection); Comparison: High-TMB versus intermediate-TMB and/or low-TMB based on cut-off values reported in the individual studies; Outcome: OS or DFS; Studies: Randomized and non-randomized prospective clinical studies, and retrospective studies. In case of duplication of data in multiple publications, the most recently published study was included.

The exclusion criteria were as follows: Studies without original data (narrative or systematic review, editorial, commentary, and case reports), and studies with incomplete data.

Data extraction and quality assessment

The following data were extracted: Name of the first author, the country where the study was performed, year of publication, number of total participants and each group (high TMB and low/intermediate-TMB), study design, NSCLC subtypes, demographic data (male and ever-smokers), stage, sample source, TMB detection method, TMB threshold, median TMB value and its range, median follow-up duration, and clinical endpoints. In addition, HR and 95% CI, source of time-to-event data, and associated adjusted variables were also reported. The HR was reported as high TMB versus low TMB.

The methodology quality of studies was assessed using the Newcastle-Ottawa Scale (NOS).²² Each study was scored under three categories: selection (representativeness of exposed cohort, selection of non-exposed cohort, and ascertainment of exposure), comparability of cohorts, and outcome (assessment of outcome, adequacy, and length of follow-up). The cumulative scores defined the risk of bias in the individual studies: 7–9 as low risk, 4–6 as moderate risk, and 0–3 as a high risk of bias.

Statistical methods and data analysis

The association between NSCLC survival and TMB was determined by pooled HR derived from the inverse variance method. The between-study heterogeneity was assessed using Cochran’s Q test and measured by the I² indices. Significant heterogeneity was considered if I² > 60%, and a random-effects model was used. In the case of moderate (I² = 30-60%) and low (I² < 30%) heterogeneity, a fixed-effects model was utilized.

Investigation into the sources of heterogeneity was performed through the publication bias assessment, sensitivity, and subgroup analyses. Funnel plot asymmetry and a significant Egger et al. ’s regression test (p < 0.1) suggested publication bias.²³ In that case, Duval and Tweedie’s trim and fill method was employed to derive an adjusted pooled estimate [adjusted HR (aHR)].²⁴ Sensitivity analyses were conducted using the leave-one-out method. Subgroup analyses based on NSCLC subtypes (LUAD and others), ethnicity (East Asian and non-Asian), adjustment of clinical covariates (multivariate and univariate/extracted data), sample size (</> 100 patients), TMB cut-off values (</> 10 Mut/Mb), and TMB assessment methods [whole exome sequencing (WES) and targeted gene panels]. A minimum of two studies should have reported clinical attributes for subgroup analysis. To reduce the chance of type I error from multiple testing on the same dataset, the Bonferroni correction was performed for the subgroup analyses.²⁵ Primary, sensitivity, and subgroup analyses were performed using Review Manager 5.4 (The Cochrane Collaboration, Copenhagen, Denmark). Publication bias analyses were performed in the JASP software (JASP 0.16, the JASP team).²⁶ The certainty of the evidence for primary outcomes was evaluated using GRADE criteria based on the number of studies, type of evidence, study quality, consistency, directness, and effect size.²⁷

Results

Literature search

Of 2975 citations following the initial database search including 76 conference abstracts, 874 duplicates were removed and 1982 articles were excluded due to irrelevant topics, absence of original data, or incomplete data. Manual reference search did not yield any new article. Subsequently, 119 articles underwent full-text screening and 10 studies finally fulfilled the inclusion criteria.^{16–19,28–33} The PRISMA flowchart for the study selection process is presented in Figure 1.

Figure 1.

PRISMA flow diagram for the study selection process.

Study characteristics and quality assessment

Ten studies comprised 2520 patients (range 55–1008) with a median sample size of 124. Table 1 features the detailed study characteristics. All studies were retrospective cohorts in design conducted in North America,^28–31,33 Europe,^18,32 and Asia.^16,17,19 Self-reported non-Hispanic-East Asian ethnicity was observed in 15% of the study population. Approximately 56% of the entire population were males and 43% were ever-smokers. Data were available on LUAD from six studies^{17,18,28,29,31,32} and LSCC from two studies.^28,33

Table 1.

Study characteristics.

Author, year	Country	Sample size (n)	Histology	Stage (n)I/II/III	Age (median, months)	Male (n,%)	Smoker (n,%)	Sample source	Diagnostic platform	Cutoff	Median TMB level (range, Mut/Mb)	Clinical endpoints	Follow-up (months)	Source of HR	Adjusted variables	NOS
Devarakonda, 2018	USA	1008	NSCLC	114/676/114	60	662 (66)	NA	Tissue	NGS (SeqCap)	4	5.7 (0.19–225)	OS, DFS, LCSS	NA	MV	Treatment, age, sex, PS, histology, type of surgery, stage	7
Yu, 2018	USA	202	SCC	NA	72.75	128 (63)	193 (85)	Tissue	NGS (GMS)	Median	31 (17–195)*	OS	NA	UV	–	6
Owada-Ozaki, 2019	Japan	92	NSCLC	58/18/15	70	63 (70)	62 (70)	Tissue	WES	Median	62 (10–502)*	OS, DFS	26 (2.8–48.3)	MV	Age, sex, smoker, histology, tumor size, stage, lymph node, EGFR, TP53, KRAS, m ERBB2, adjuvant therapy	9
Jiang, 2019	China	189	SCC	92/54/43	63	179 (94)	129 (68)	Tissue	WES	Median	9.43	OS, DFS	NA	MV	Age, Sex, Smoking, PS, Stage, PI, VI, tumor differentiation, PD-L1 status, CD8 + TIL	7
Mitchell, 2019	USA	55	NSCLC	33/19/3	67	25 (45)	47 (85)	Tissue	WES	Median	3.91 (1.59–8.71)	DFS	16.1 (10.1–22.6)	Survival curve	–	5
Wang, 2019	China	101	LUAD	19/18/55	59	62 (61)	37 (36)	NA	WES	Mean	163.5*	DFS	NA	Survival curve	–	5
Talvitle, 2020	Finalnad	148	LUAD	95/38/37	66	95 (64)	137 (92)	Tissue	NGS (F1 CDx)	14	7.02	OS, LSCC	45 (0.6–167.6)	MV	Stage, histological grade, adjuvant therapy	8
Jones, 2021	USA	604	LUAD	447/95/62	68	202 (33)	466 (77)	Tissue	NGS (MSK IMPACT)	13.8	NA	DFS	30.1 (20.9–39.6)	UV	–	8
Ruel, 2021	Canada	92	NSCLC	58/18/5	66	17 (18)	0	Tissue	NGS (F1 CDx)	Median	1.57 (0.13–17.94)	OS, DFS	87	UV	–	8
Bubendorf, 2022	Switzerland	78	LUAD	43/35/NI	66	48 (61)	63 (81)	Tissue	NGS (F1 CDx)	10	7.6 (5–16.1)	OS, DFS, TTR	56.1 (38.8–72)	MV	Gender, Age at surgery, Adjuvant chemotherapy, h/o cancer, Stage, Localization of primary tumor, Tumor size, Year of surgery, surgery technique and anatomy	9

DFS, disease-free survival; F1 CDX, FoundatioOne, CDx™; GMS, genomic modeling system; h/o, history of; LUAD, Adenocarcinoma; LSCC, lung cancer-specific survival; Mut/Mb, Mutations per Megabase; MV, multivariate; NA, not available; NGS, Next generation sequencing; NOS, Newcastle-Ottawa Scale; OS, overall survival; PI, pleural invasion; PS, performance status; SCC, squamous cell carcinoma; TTR, time to response; UV, univariate; VI, vascular invasion; WES, whole exome sequencing.

The sample source was reported in all except one study,¹⁷ and most studies used Formalin-Fixed Paraffin-Embedded (FFPE) resected primary tumor specimens. WES (included in survival analysis) was employed in four studies,^16,17,19,30 whereas the rest of the studies used targeted gene panels for TMB assessment. Most studies divided the comparative groups into high-TMB and low-TMB based on TMB cut-off values. Devarakonda et al. divided the patients into three groups, namely, high-TMB (⩾8 Mut/Mb), intermediate-TMB (4–8 Mut/Mb), and low-TMB (⩽4 Mut/Mb), of which we included high- versus moderate-TMB for survival analyses.²⁸ The TMB cutoff values varied considerably between individual studies and have been reported in detail in Table 1.

Stages I and II had similar distribution, seen in 38% of patients. Median follow-up ranged between 16.1 and 87 months. Data on OS and DFS were available from seven,^{16,18,19,28,31–33} and nine studies,^{16–19,28,30–32} respectively. Five studies evaluated the effects of TMB level on survival outcome through multivariate Cox proportional hazards regression models, adjusting for clinicopathological variables, such as age, gender, smoking status, stage, adjuvant therapy, histological subtypes, and histological grades.^{16,18,19,28,32} Furthermore, Owada-Ozaki et al. incorporated TP53, EGFR, KRAS, and ERBB2 mutations in this model.¹⁶ Two studies required the extraction of data from the survival curves.^17,30

The results from the NOS revealed that seven studies had a low risk^{16,18,19,28,29,31,32} and the rest had a moderate risk of bias (Table 1 and Supplemental Table S3). Most studies provided independent and comparable cohorts, adequate ascertainment of exposure, appropriate statistical analysis, and thorough experimental results. A lack of data on the adequacy of follow-up and on sufficient follow-up time for the event to transpire were the major deficiencies in a few studies.^17,19,28,33 The median and the mean score was 7 (range 5–9), suggesting an overall high methodological quality.

Pooled effects of HR for OS

From seven studies and 1760 patients, the pooled HR for OS was 1.18 (95% CI, 0.70, 1.33; p 0.53), indicating a lack of significant survival benefits between high-TMB and low-TMB groups. A significant between-study heterogeneity was observed (I² = 80%, p_het <0.0001; Figure 2).

Figure 2.

Forest plot of HR for OS in patients with high TMB versus low TMB.

The subgroup analysis results are presented in Table 2 and Supplemental Figures S1, S2, S3, S4, and S5. None of the subgroups’ pooled effects showed a statistically significant association between TMB and OS. All except NSCLC subtypes subgroup analyses showed significant between-study heterogeneity. For NSCLC subtypes subgroup analysis, the LUAD subgroup [I² = 85%, p_het < 0.0001] had a significant heterogeneity whereas LSCC did not [I² = 0%, p_het = 0.48]. Due to a limited number of studies, TMB cutoff value subgroup analysis could not be carried out.

Table 2.

Summary of the meta-analysis results.

Outcomes	Subgroups	Pooled HR	p Value	Heterogeneity		Subgroup difference
Outcomes	Subgroups	Pooled HR	p Value	P	p Value	Subgroup difference
Overall survival	Primary analysis	1.18 [0.70, 1.99]	0.53	80	<0.0001
	Ethnicity
	Non-Asian	0.99 [0.54, 1.84]	0.99	82	0.0002	0.29
	East Asian	3.21 [0.39, 26.25]	0.28	75	0.05
	Sample size
	<100	2.70 [0.78, 9.41]	0.12	75	0.02	0.09
	>100	0.85 [0.51, 1.41]	0.53	77	0.004
	TMB assessment
	Targeted-panel	0.99 [0.54, 1.84]	0.99	82	0.0002	0.58
	WES	3.21 [0.39, 26.25]	0.28	75	0.05
	TMB cut-off values
	<10 Mut/Mb	1.38 [0.82, 2.30]	0.22	72	0.01	0.29
	>10 Mut/Mb	1.13 [0.26, 4.83]	0.87	87	0.0004
	Survival analysis methods
	Multivariate	0.96 [0.52, 1.78]	0.89	78	0.0009	0.29
	Univariate/extracted data	1.92 [0.61, 6.03]	0.26	83	0.02
	NSCLC subtypes
	Adenocarcinoma	1.01 [0.42, 2.43]	0.99	85	0.0001	0.58
	Squamous cell carcinoma	1.31 [0.97, 1.77]	0.08	0	0.48
Disease-free survival
	Primary analysis	1.18 [0.91, 1.52]	0.20	75	0.0001
	Ethnicity
	Non-Asian	1.04 [0.78, 1.38]	0.81	76	0.001	0.02
	East Asian	1.68 [1.24, 2.26]	0.0007	0		0.86
	Sample size
	<100	1.59 [1.17, 2.17]	0.003	0	0.39	0.03
	>100	1.00 [0.74, 1.35]	0.99	80	0.0004
	TMB assessment
	Targeted-panel	0.95 [0.68, 1.32]	0.75	76	0.002	0.01
	WES	1.59 [1.26, 2.00]	0.0001	0	0.89
	TMB cut-off values
	<10 Mut/Mb	1.31 [0.86, 1.98]	0.20	62	0.03	0.24
	>10 Mut/Mb	1.04 [0.62, 1.76]	0.88	85	0.00001
	Survival analysis methods
	Multivariate	0.92 [0.50, 1.69]	0.79	68	0.01	0.22
	Univariate/extracted data	1.44 [0.99, 2.10]	0.06	82	0.0007
	NSCLC subtypes
	Adenocarcinoma	0.97 [0.67, 1.40]	0.85	75	0.003

NSCLC, non-small cell lung cancer; TMB, Tumor mutation burden; WES, Whole exome sequencing.Bold values suggest a statistically significant effect estimate.

Pooled effects of HR for DFS

From nine studies and 2365 patients, the pooled HR for DFS was 1.18 (95% CI, 0.91, 1.52; p = 0.53), suggesting no significant survival difference between the high-TMB and low-TMB groups. Significant heterogeneity was observed (I² = 75%, p_het = 0.0001; Figure 3).

Figure 3.

Forest plot of HR for DFS in patients with high TMB versus low TMB.

The subgroup analysis results are presented in Table 2. The results of the subgroup analyses indicated that ethnicity (I² = 80.8%, p_subgroup = 0.02), sample size (I² = 78%, p_subgroup = 0.03), and TMB assessment method (I² = 83.8%, p_subgroup = 0.01) significantly modified the association of TMB with DFS. In the East Asian subgroup, the DFS of patients with high TMB was shorter than patients with low TMB (HR, 1.68, 95% CI 1.24, 2.26; p = 0.0007; Supplemental Figure S6). For the sample size subgroup, when the sample size was fewer than or equal to 100 participants, the DFS of patients with high TMB was shorter than that of low TMB (HR, 1.59, 95% CI 1.17, 2.17; p = 0.003; Supplemental Figure S7. In the WES subgroup, DFS was shorter in patients with high TMB levels (HR, 1.59, 95% CI 1.26, 2.00; p = 0.0001), whereas DFS was similar between high- and low-TMB groups in the targeted-gene panel subgroup (Supplemental Figure S8). Meanwhile, the heterogeneity of the East Asians, the sample size of fewer than 100 patients, and the WES subgroup were resolved. The association of DFS and TMB level was unrelated to survival analysis methods (Supplemental Figure S9) and the TMB cut-off values (Supplemental Figure S10). Due to a limited number of studies, NSCLC subtypes subgroup analysis could not be performed. However, additional analysis of studies with LUAD exclusively showed a consistent result as the primary analysis (HR, 0.97, 95% CI 0.67, 1.40; p = 0.85; I² = 75%, p_het = 0.003; Supplemental Figure S11).

Sensitivity and publication bias

The leave-one-out method exhibited non-statistically significant variation in the pooled OS and DFS effects. The presence of symmetrical funnel plots and Egger’s regression tests [OS, p = 0.107; DFS = 0.828] indicated a lack of publication bias (Supplemental Figure S12 and S13). Based on the GRADE criteria, the certainty of the evidence for OS and DFS was low (Supplemental Table S4).

Discussion

Our meta-analysis evaluated the hypothesis of whether TMB was associated with early-stage, resected NSCLC survival. Data from 10 studies comprising 2520 patients suggested that patients with high TMB have a similar OS and DFS compared to low TMB. Furthermore, primary analyses were associated with significant between-study heterogeneity. Subgroup analysis for OS indicated that NSCLC subtypes might explain high heterogeneity. Furthermore, subgroup analyses for DFS revealed that ethnicity, sample size, and the TMB assessment methods might explain significant between-study heterogeneity. In addition, patients of East Asian ethnicity, studies with fewer than 100 patients, and the use of WES were associated with poor DFS in the high TMB group. Overall, the GRADE criteria indicated a low certainty of the evidence for both primary analyses’ outcomes.

Prior studies have shown that high TMB was associated with improved OS and PFS following ICI therapy in melanoma, colorectal cancer, and NSCLC patients.^34–36 Examination of the prognostic role of TMB assessed using MSK-IMPACT in a large, advanced solid cancer patient’s cohort who did not receive ICIs, indicated a lack of association between high-TMB (highest mutation load quintile 20%) and OS (HR, 1.12, p = 0.11).³⁷ A similar lack of prognostic impact was observed in metastatic NSCLC (n = 623; HR 1.108, p = 0.508).³⁷ Similarly, a pooled analysis of seven studies on advanced solid tumors not treated with ICIs showed no survival difference between high and low TMB.³⁸ In early-stage NSCLC, high TMB seems to have no general prognostic benefit per our findings. This lack of prognostic benefit was also observed within both NSCLC subtypes.

Being a genetic biomarker, ethnic or racial differences in TMB appear intuitive.^39,40 In a continuously enrolled cohort of advanced NSCLC patients, patients of African genomic ancestry had the highest median TMB level (8.75 Mut/Mb). In contrast, the patients of Asian ancestry had the lowest median TMB level (3.75 Mut/Mb), despite stratification for smoking status (p < 0.001).⁴¹ One of the key findings in our study was the presence of ethnicity-specific differences in DFS outcome by TMB, where patients of East Asian ethnicity had worse DFS in the high-TMB group (p = 0.0007). This finding was consistent with an analysis of the non-ICI treated patients in the Cancer Genome Atlas (TCGA) and MSK-IMPACT cohorts, where high-TMB (>10 Mut/Mb) negatively affected survival in Asian patients (HR 2.99; 95% CI 1.22, 4.39).⁴² The ethnicity-specific association between TMB and clinical outcome is likely due to (1) the lack of statistical power to detect subgroup differences due to unequal covariate distribution in the subgroups (East Asian: 3 studies, 380 participants; non-East Asian: 6 studies, 1985 participants), (2) residual analytical issues in TMB assessment: Five out of six studies in the non-Asian subgroup used tumor-only panels which are shown to have ‘TMB-inflation’, particularly in patients of Asian or African ancestry than patients of European ancestry,^{18,28,29,31,32,43} (3) implications of social healthcare determinants, including differential smoking habits and healthcare disparities.^44,45 However, considering comparable healthcare access between Asians and non-Hispanic Whites and a similar mutational profile between Asian and Black patients, the race-related genetic and environmental factors have a multi-faceted relationship with TMB and thus, require further exploration.^46,47

Targeted gene panel assays are increasingly being considered in clinical practice compared to WES because of their short turnaround time, identification of targetable mutation, and cost-effectiveness, despite the latter being the current reference standard.⁴⁸ Numerous preanalytical factors such as specimen types, quality, and quantity; sequencing factors such as gene list, panel size, and sequencing depth; and bioinformatics algorithms with somatic and germline variant calling and filtering are likely to affect TMB estimates.^49,50 These factors may also explain the targeted-gene panel subgroup as one of the main sources of heterogeneity in our review.

Based on the mounting evidence on the prognostic role of TMB assessed in circulating tumor DNA, B-FIRST (NCT02848651), a phase II clinical trial, revealed that advanced NSCLC patients (IIIB-IV) who received first-line atezolizumab had longer OS and higher ORR if blood TMB (bTMB) was ⩾16 (14.5 Mut/Mb).^51–53 Nine studies in our review reported the use of FFPE tumor specimens which may lead to TMB overestimation due to deamination from formalin fixation.^{16,18,19,28–33,54} Only one study reported tumor fractionation and even low tumor fractions might result in deflated TMB levels.^29,55 Four studies utilized the U.S. Food and Drug Agency approved gene panel tests, namely, FoundationOne, FoundationOne CDx (F1 CDx), and MSK-IMPACT, that have shown acceptable concordance with WES in empirical and in-silico TMB assessment.^{18,29,31,32,56,57} Similarly, a pooled analysis from TCGA database did not find a significant prognostic impact of TMB on LUAD (HR 1.35, 95% CI 0.89, 2.05) and LSCC (HR, 0.77, 95% CI 0.50, 1.18) survival independent of TMB assessment methods (WES, MSK-IMPACT, and FoundationOne CDx).⁵⁸ Two international consortiums, the Friends of Cancer Research and Quality Assurance Initiative Pathology validated bioinformatic algorithms, standardized panel-based TMB estimates using in-silico analysis, and assessed between-TMB panel variations in FFPE tumor specimens for multiple cancer types, thereby taking TMB one step closer to a more refined prognostic biomarker.^59–61

Evaluation of TMB, assessed using F1 CDx assay, as a predictive biomarker for ICI response in numerous landmark clinical trials on NSCLC utilized ⩾10 mut/Mb (~200 mutations by WES) as the cut-off for defining ‘high TMB’.^52,62–64 A 23-gene TMB estimation panel evaluated in an early-stage, resected NSCLC Chinese cohort (n = 89) showed a statistically significant association between DFS and TMB only when TMB cut-offs were between two and four mutation counts, despite having a high concordance with WES (Spearman r = 0.8487, p < 0.0001).⁶⁵ TMB threshold for prognostic purposes depends on the tumor microenvironment and may vary according to the cancer types, the assay, and clinical factors, like race and treatment history.^43,66 Six studies in our review used either tertile cut-offs or median TMB levels contributing to high heterogeneity.^{16,19,28,30,31,33} Due to the lack of comparability, these TMB cut-offs cannot be used in future clinical studies.⁶⁷ The phase III of the FoCR study is aimed at validating TMB thresholds for individual cancer types thereby allowing cross-trial comparisons and eventually its widespread clinical implementation.⁶⁸

Several aspects of the present study strengthen its role in the contemporary NSCLC therapeutic landscape. Our meta-analysis is the first to report the pooled effect estimates of OS and DFS in early-stage, resected NSCLC patients by TMB levels. The lack of inherent prognostic value of TMB in this large, non-ICI-treated cohort provides a benchmark for future studies, considering the accumulating evidence of the role of neoadjuvant immunotherapy in early-stage NSCLC.^69–71

Despite adhering to delivering a statistically robust meta-analysis, certain limitations prevailed. Although two studies in the OS and three studies in the DFS analyses showed a prognostic association of high TMB in early-stage NSCLC, the limited sample sizes (ranging from 55 to 148) in those individual studies portend limited statistical power. There is substantial evidence supporting this contention, as the association between NSCLC DFS and TMB levels is significantly stronger in cohorts with fewer than 100 participants (p = 0.003). Although publication bias in the OS and DFS was not evident, Egger’s regression or symmetry of funnel plots is not reliable in a meta-analysis with fewer than 10 studies.^72,73 Lastly, variable study designs, multiple TMB assessment methods, unadjusted clinicopathological characteristics, and the different survival analysis methods may further confer heterogeneity. However, we strived to ascertain the sources of heterogeneity through subgroup analyses. Based on GRADE criteria, the true effect might be markedly different from the estimated effect, thus requiring more research into this area.⁷⁴

Conclusion

Our current systematic review and meta-analysis provide evidence against an inherent prognostic role of TMB in early-stage, resected NSCLC patients. In the East-Asian patients, high TMB was associated with shorter DFS; nonetheless, the association is not apparent in non-Asian patients. Ethnicity, TMB assessment methods, and sample size were the sources of between-study heterogeneity for DFS. Overall, as per the GRADE criteria, the certainty of the evidence is low and thus, further research in harmonization and standardization of TMB assays is required for its widespread utilization in clinical practice.

Supplemental Material

sj-docx-1-tam-10.1177_17588359231195199 – Supplemental material for The prognostic value of TMB in early-stage non-small cell lung cancer: a systematic review and meta-analysis

Supplemental material, sj-docx-1-tam-10.1177_17588359231195199 for The prognostic value of TMB in early-stage non-small cell lung cancer: a systematic review and meta-analysis by Durgesh Wankhede, Sandeep Grover and Paul Hofman in Therapeutic Advances in Medical Oncology

Footnotes

Acknowledgements

None.

Declarations

ORCID iD

Durgesh Wankhede

Supplemental material

Supplemental material for this article is available online.

References

Herbst

Morgensztern

Boshoff

. The biology and management of non-small cell lung cancer. Nature 2018; 553: 446–454.

National Cancer Institute. SEER cancer stat facts: lung and bronchus cancer. National Cancer Institute, Bethesda, MD, https://seer.cancer.gov/statfacts/html/lungb.html (2021, accessed 11 January 2023).

Chansky

Detterbeck

Nicholson

, et al. The IASLC lung cancer staging project: external validation of the revision of the TNM stage groupings in the eighth edition of the TNM classification of lung cancer. J Thorac Oncol 2017; 12: 1109–1121.

Wankhede

. Evaluation of Eighth AJCC TNM Sage for lung cancer NSCLC: a meta-analysis. Ann Surg Oncol 2021; 28: 142–147.

Song

Wang

, et al. A combined Two-mRNA signature associated with PD-L1 and tumor mutational burden for prognosis of lung adenocarcinoma. Front Cell Dev Biol 2021; 9: 634697.

Nghiem

Bhatia

Lipson

, et al. PD-1 blockade with pembrolizumab in advanced Merkel-cell carcinoma. N Engl J Med 2016; 374: 2542–2552.

Schumacher

Schreiber

. Neoantigens in cancer immunotherapy. Science 2015; 348: 69–74.

Chan

Yarchoan

Jaffee

, et al. Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol 2019; 30: 44–56.

Rizvi

Hellmann

Snyder

, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015; 348: 124–128.

10.

Goodman

Kato

Bazhenova

, et al. Tumor mutational burden as an independent predictor of response to immunotherapy in diverse cancers. Mol Cancer Ther 2017; 16: 2598–2608.

11.

Hellmann

Nathanson

Rizvi

, et al. Genomic features of response to combination immunotherapy in patients with advanced non-small-cell lung cancer. Cancer Cell 2018; 33: 843–852.e4.

12.

Rizvi

Sanchez-Vega

, et al. 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.

13.

Galvano

Gristina

Malapelle

, et al. The prognostic impact of tumor mutational burden (TMB) in the first-line management of advanced non-oncogene addicted non-small-cell lung cancer (NSCLC): a systematic review and meta-analysis of randomized controlled trials. ESMO Open 2021; 6: 100124.

14.

Deng

Zhao

Cai

, et al. PD-L1 expression and tumor mutation burden as pathological response biomarkers of neoadjuvant immunotherapy for early-stage non-small cell lung cancer: a systematic review and meta-analysis. Crit Rev Oncol Hematol 2022; 170: 103582.

15.

Talvitie

E-M

Liljeroos

Vilhonen

, et al. Comprehensive genomic profiling of Finnish lung adenocarcinoma cohort reveals high clinical actionability and SMARCA4 altered tumors with variable histology and poor prognosis. Neoplasia 2022; 32: 100832.

16.

Owada-Ozaki

Muto

Takagi

, et al. Prognostic impact of tumor mutation burden in patients with completely resected Non-Small cell lung cancer: Brief Report. J Thorac Oncol 2018; 13: 1217–1221.

17.

Wang

Liang

Lin

, et al. Molecular subtyping and prognostic assessment based on tumor mutation burden in patients with lung adenocarcinomas. Int J Mol Sci 2019; 20: 4251.

18.

Bubendorf

Zoche

Dafni

, et al. Prognostic impact of tumour mutational burden in resected stage I and II lung adenocarcinomas from a European Thoracic Oncology Platform Lungscape cohort. Lung Cancer 2022; 174: 27–35.

19.

Jiang

Shi

Dong

, et al. Genomic landscape and its correlations with tumor mutational burden, PD-L1 expression, and immune cells infiltration in Chinese lung squamous cell carcinoma. J Hematol Oncol 2019; 12: 75–13.

20.

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 74: 790–799.

21.

Tierney

Stewart

Ghersi

, et al. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials 2007; 8: 16.

22.

Wells

Shea

O’Connell

, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses, http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp (2013, accessed 2 September 2021).

23.

Egger

Davey Smith

Schneider

, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–634.

24.

Duval

Tweedie

. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 2000; 56: 455–463.

25.

Armstrong

. When to use the Bonferroni correction. Ophthalmic Physiol Opt 2014; 34: 502–508.

26.

JASP Team. JASP. Published online. 2022. www.jasp-stats.org

27.

Guyatt

Oxman

Akl

, et al. GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin Epidemiol 2011; 64: 383–394.

28.

Devarakonda

Rotolo

Tsao

M-S

, et al. Tumor mutation burden as a biomarker in resected non-small-cell lung cancer. J Clin Oncol 2018; 36: 2995–3006.

29.

Jones

Caso

Tan

, et al. KRASG12C mutation is associated with increased risk of recurrence in surgically resected lung AdenocarcinomaKRASG12C and genomic correlates in lung adenocarcinoma. Clin Cancer Rev 2021; 27: 2604–2612.

30.

Mitchell

Negrao

Parra

, et al. Lymphovascular invasion is associated with mutational burden and PD-L1 in resected lung cancer. Ann Thorac Surg 2020; 109: 358–366.

31.

Ruel

L-J

Gaudreault

, et al. Tumor mutational burden by whole-genome sequencing in resected NSCLC of never smokers. Cancer Epidemiol Biomarkers Prev 2022; 31: 2219–2227.

32.

Talvitie

E-M

Vilhonen

Kurki

, et al. High tumor mutation burden predicts favorable outcome among patients with aggressive histological subtypes of lung adenocarcinoma: a population-based single-institution study. Neoplasia 2020; 22: 333–342.

33.

Chen

Ballman

, et al. Correlation of PD-L1 expression with tumor mutation burden and gene signatures for prognosis in early-stage squamous cell lung carcinoma. J Thorac Oncol 2019; 14: 25–36.

34.

Ning

Liu

Wang

, et al. The predictive value of tumor mutation burden on clinical efficacy of immune checkpoint inhibitors in melanoma: a systematic review and meta-analysis. Front Pharmacol 2022; 13: 748674.

35.

, et al. Tumor mutational burden predicting the efficacy of immune checkpoint inhibitors in colorectal cancer: a systematic review and meta-analysis. Front Immunol 2021; 12. DOI: 10.3389/fimmu.2021.751407

36.

Yarchoan

Hopkins

Jaffee

. Tumor mutational burden and response rate to PD-1 inhibition. N Engl J Med 2017; 377: 2500–2501.

37.

Samstein

Lee

C-H

Shoushtari

, et al. Tumor mutational load predicts survival after immunotherapy across multiple cancer types. Nat Genet 2019; 51: 202–206.

38.

Cao

, et al. High tumor mutation burden predicts better efficacy of immunotherapy: a pooled analysis of 103078 cancer patients. OncoImmunology 2019; 8: e1629258.

39.

Choudhury

Eghtesad

Kadri

, et al. Fewer actionable mutations but higher tumor mutational burden characterizes NSCLC in black patients at an urban academic medical center. Oncotarget. 2019; 10: 5817–5823.

40.

Heath

Lynce

Xiu

, et al. Racial disparities in the molecular landscape of Cancer. Anticancer Res 2018; 38: 2235–2240.

41.

Brawley

Luhn

Reese-White

, et al. Disparities in tumor mutational burden, immunotherapy use, and outcomes based on genomic ancestry in non-small-cell lung cancer. J Glob Oncol 2021; 7: 1537–1546.

42.

Hsiehchen

Espinoza

Valero

, et al. Impact of tumor mutational burden on checkpoint inhibitor drug eligibility and outcomes across racial groups. J Immunother Cancer 2021; 9: e003683.

43.

Nassar

Adib

Abou Alaiwi

, et al. Ancestry-driven recalibration of tumor mutational burden and disparate clinical outcomes in response to immune checkpoint inhibitors. Cancer Cell 2022; 40: 1161–1172.e5.

44.

Tannenbaum

Koru-Sengul

Zhao

, et al. Survival disparities in non-small cell lung cancer by race, ethnicity, and socioeconomic status. Cancer 2014; 20: 237–245.

45.

Cheng

Cramb

Baade

, et al. The International Epidemiology of lung cancer: latest trends, disparities, and tumor characteristics. J Thorac Oncol 2016; 11: 1653–1671.

46.

Trinh

Q-D

Nguyen

Leow

, et al. Cancer-specific mortality of Asian Americans diagnosed with cancer: a nationwide population-based assessment. J Natl Cancer Inst 2015; 107: djv054–djv054.

47.

Zhang

Edwards

Flemington

, et al. Racial disparities in patient survival and tumor mutation burden, and the association between tumor mutation burden and cancer incidence rate. Sci Rep 2017; 7: 13639.

48.

Vokes

Liu

Ricciuti

, et al. Harmonization of tumor mutational burden quantification and association with response to immune checkpoint blockade in non-small-cell lung cancer. JCO precis oncol 2019; 3: 1–12.

49.

Büttner

Longshore

López-Ríos

, et al. Implementing TMB measurement in clinical practice: considerations on assay requirements. ESMO Open 2019; 4: e000442.

50.

Endris

Buchhalter

Allgäuer

, et al. Measurement of tumor mutational burden (TMB) in routine molecular diagnostics: in silico and real-life analysis of three larger gene panels. Int J Cancer 2019; 144: 2303–2312.

51.

Herbst

Giaccone

de Marinis

, et al. Atezolizumab for first-line treatment of PD-L1-Selected patients with NSCLC. N Engl J Med 2020; 383: 1328–1339.

52.

Rizvi

Cho

Reinmuth

, et al. Durvalumab with or without tremelimumab vs platinum-based chemotherapy as first-line treatment for metastatic non-small cell lung cancer: MYSTIC. Ann Oncol 2018; 29: x40–x41.

53.

Kim

Velcheti

Mekhail

, et al. Blood-based tumor mutational burden as a biomarker for atezolizumab in non-small cell lung cancer: The phase 2 B-F1RST trial. Nat Med 2022; 28: 939–945.

54.

Chen

Luthra

Goswami

, et al. Analysis of pre-analytic factors affecting the success of clinical next-generation sequencing of solid organ malignancies. Cancers 2015; 7: 1699–1715.

55.

Srinivasan

Sedmak

Jewell

. Effect of fixatives and tissue processing on the content and integrity of nucleic acids. Am J Pathol 2002; 161: 1961–1971.

56.

Johnson

Frampton

Rioth

, et al. Targeted next generation sequencing identifies markers of response to PD-1 blockade. Cancer Immunol Res 2016; 4: 959–967.

57.

Garofalo

Sholl

Reardon

, et al. The impact of tumor profiling approaches and genomic data strategies for cancer precision medicine. Genome Med 2016; 8: 79–10.

58.

H-x

Wang

Z-X

Zhao

, et al. Tumor mutational and indel burden: a systematic pan-cancer evaluation as prognostic biomarkers. Ann Transl Med 2019; 7: 640.

59.

Merino

McShane

Fabrizio

, et al. Establishing guidelines to harmonize tumor mutational burden (TMB): in silico assessment of variation in TMB quantification across diagnostic platforms: phase I of the Friends of Cancer Research TMB Harmonization Project. J Immunother Cancer 2020; 8: e000147.

60.

Vega

Yee

McShane

, et al. Aligning tumor mutational burden (TMB) quantification across diagnostic platforms: phase II of the Friends of Cancer Research TMB Harmonization Project. Ann Oncol 2021; 32: 1626–1636.

61.

Stenzinger

Allen

Maas

, et al. Tumor mutational burden standardization initiatives: recommendations for consistent tumor mutational burden assessment in clinical samples to guide immunotherapy treatment decisions. Genes Chromosomes Cancer 2019; 58: 578–588.

62.

Hellmann

Ciuleanu

T-E

Pluzanski

, et al. Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. N Engl J Med 2018; 378: 2093–2104.

63.

Paz-Ares

Ciuleanu

T-E

Cobo

, et al. First-line nivolumab plus ipilimumab combined with two cycles of chemotherapy in patients with non-small-cell lung cancer (CheckMate 9LA): an international, randomised, open-label, phase 3 trial. Lancet Oncol 2021; 22: 198–211.

64.

Fehrenbacher

Spira

Ballinger

, et al. Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial. Lancet 2016; 387: 1837–1846.

65.

Tian

Chu

, et al. A novel tumor mutational burden estimation model as a predictive and prognostic biomarker in NSCLC patients. BMC Med 2020; 18: 232.

66.

Sha

Jin

Budczies

, et al. Tumor mutational burden as a predictive biomarker in solid tumors. Cancer Discov 2020; 10: 1808–1825.

67.

Addeo

Friedlaender

Banna

, et al. TMB or not TMB as a biomarker: that is the question. Crit Rev Oncol Hematol 2021; 163: 103374.

68.

Sung

M-T

Wang

Y-H

. Open the technical black box of tumor mutational burden (TMB): factors affecting harmonization and standardization of panel-based TMB. Int J Mol Sci 2022; 23: 5097.

69.

Forde

Chaft

Smith

, et al. Neoadjuvant PD-1 blockade in resectable lung cancer. N Engl J Med 2018; 378: 1976–1986.

70.

Cascone

William

Weissferdt

, et al. Neoadjuvant nivolumab or nivolumab plus ipilimumab in operable non-small cell lung cancer: the phase 2 randomized NEOSTAR trial. Nat Med 2021; 27: 504–512.

71.

Altorki

Borczuk

Saxena

, et al. P2.04-92 neoadjuvant durvalumab with or without sub-ablative stereotactic radiotherapy (SBRT) in patients with resectable NSCLC (NCT02904954). J Thorac Oncol 2019; 14: S746.

72.

Schwarzer

Carpenter

Rücker

, et al. Small-study effects in meta-analysis. In: Schwarzer

(ed.) Meta-analysis with R. Cham: Springer, 2015, pp. 107–141.

73.

Dalton

Bolen

Mascha

. Publication bias: the elephant in the review. Anesth Analg 2016; 123: 812–813.

74.

Guyatt

Oxman

Vist

, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008; 336: 924–926.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

10.59 MB