Could TMB be a Predictor of Immunotherapy Response in NSCLC?

By: Shweta Srivastava and Anil Sharma

Tumor mutational burden (TMB (the number of mutations per megabase of DNA (mut/Mb)) can be a predictor of response for single-agent immunotherapy, but it has failed to demonstrate any correlation with response when immunotherapy is combined with chemotherapy. The benefit of the combination can be maintained independent of TMB status, and evaluation of TMB is not required for making treatment decisions for patients eligible to receive combination therapy.

Chemotherapy increases antigenicity via immunogenic cell death, helping transform the tumor microenvironment (TME) to T-cell-inflamed (immunologically hot), rendering it responsive to PD-1/PD-L1 treatment. This could be one of the reasons why immunotherapies are the treatment of choice in 2L+ patients progressing on chemotherapy. This also suggests that an early on-treatment measurement for TMB is more informative than a pre-treatment evaluation. It may help in selecting patients progressing from previous settings (adjuvant or 1L) after receiving chemotherapy who will benefit most from immunotherapy treatment. This indicates that TMB has the potential to predict responses if patients are already chemotherapy-exposed and now eligible for immunotherapy treatment.

Although multiple trials fail to show a correlation between TMB and PD-L1 expression, greater benefit is consistently observed with a single anti-PD-1/PD-L1 agent in patients with high TMB and PD-L1 expression, suggesting these independent biomarkers can be used in concert to select patients for immunotherapy response.

Key challenges limiting the utilization of TMB include an unclear definition of high TMB, variation in the diagnostic tests used for TMB measurement and the availability of tissue for biopsy, indicating a need for calibration and harmonization for optimal alignment on the threshold for high TMB.

The selection of low-TMB patients for immunotherapy treatment will be an area to evaluate for the validation of novel biomarkers, such as tissue DNA (tDNA), tumor-infiltrating lymphocytes (TILs), T-regulatory cells/myeloid-derived suppressor cells (Tregs/MDSCs), etc.

Analyses of key clinical trials presented at ESMO 2019 add to the confusion

Exploratory analyses of updated data from KEYNOTE-021, -189 and -407 showed no significant association between tTMB (TMB estimated from circulating tumor DNA (ctDNA) found in plasma) at a cutoff point of 175 mut/exome and efficacy of pembrolizumab plus platinum-based chemotherapy or chemotherapy alone. Similar overall survival (OS), progression-free survival (PFS) and overall response rate (ORR) benefits were reported in both the tTMB-high and tTMB-low subgroups in the pembrolizumab plus chemotherapy arm, indicating the limited ability of tTMB to discriminate between responder and non-responder patients.[1]

Conversely, exploratory analysis of KEYNOTE-010 and -042 showed tTMB at a cutoff point of 175 mut/exome was associated with efficacy outcomes (OS, PFS and ORR) for pembrolizumab but not with chemotherapy. No association was observed between tTMB and PD-L1 expression. Clinical utility of OS and PFS was reported for patients with tTMB ≥175 mut/exome but not for patients with a cutoff less than 175 mut/exome. This suggests that higher tTMB levels are associated with an improvement in clinical outcomes with pembrolizumab monotherapy in PD-L1-positive advanced NSCLC.[2]

Final efficacy results were also presented from the prospective Phase II B-F1RST trial evaluating the association of bTMB (TMB estimated by ctDNA in blood) with clinical outcomes for atezolizumab in 1L NSCLC. At a prespecified cutoff for bTMB of 16 mut/Mb, ORR was significantly greater in the bTMB-high subgroup (36.4%) than in the bTMB-low subgroup (6.4%). Although the PFS hazard ratio (HR) for mPFS looked impressive at 0.51, it failed to reach statistical significance, with a p-value of 0.1315.[3]

Rationale to evaluate TMB as a predictive biomarker for response

PD-L1 ≥50% enriches for response, but its value as a selection marker is debatable: Early studies identified PD-L1 as a stratification factor that enriched for tumor response. In NSCLC, KEYNOTE-024 showed pembrolizumab monotherapy improved ORR and prolonged PFS and OS relative to chemotherapy in treatment-naïve patients with PD-L1 expression ≥50%.[4] In contrast, CheckMate-026 (nivolumab monotherapy) failed to demonstrate benefit in patients with a broader range of PD-L1 expression (PD-L1 ≥5%).[5] The association of response with a high PD-L1 expression level in NSCLC was confirmed in KEYNOTE-042. This trial showed the response to pembrolizumab compared to chemotherapy was greater in patients with PD-L1 ≥50% (OS HR 0.69, ORR 39.5%) than in patients with PD-L1 ≥1% (OS HR 0.81, ORR 27.3%). More significantly, in patients with low PD-L1 (1-49%), pembrolizumab failed to beat chemotherapy.[6] Results from the Phase III IMpower110 study showed a similar trend, with atezolizumab beating chemotherapy in PD-L1-high patients (OS HR 0.59, p=0.0106) but not in all-comers (HR 0.83, p=0.1481).[7]

However, while this establishes high PD-L1 as an enrichment marker, a PD-L1 ≥50% cutoff excludes the majority of patients, many of whom might benefit from PD-(L)1 therapy or combinations. On the other hand, PD-L1 ≥1% captures patients who gain no benefit from PD-(L)1 inhibitors or may not do as well as when on chemotherapy. Moreover, though PD-L1 has served as an enrichment factor for response in clinical trials, its ability to identify patients likely to respond to checkpoint inhibitors is debatable. This has led to a pressing need for and a surge of interest in other predictive biomarkers that can identify patients who do not benefit from immunotherapies.

The rationale behind TMB as a biomarker for response to immunotherapy stems from the observation that high mutation load correlates with an immunogenic TME, and increased expression of tumor-specific neoantigens enables immunotherapies to mount an immune response.

Interest in TMB as a potential biomarker for response arose from an analysis of the CheckMate-026 outcomes, which revealed an association between nivolumab responses and TMB but not PD-L1 ≥5%. Among patients with a high TMB, the response rate was higher in the nivolumab group than in the chemotherapy group (47% vs. 28%), and median PFS was longer (9.7 vs. 5.8 months, HR 0.62); OS, however, was similar between the groups regardless of TMB, and exploratory analysis revealed there was no significant association between TMB and PD-L1 expression levels. Response rates and PFS in the nivolumab group were greatest in patients with both high TMB and high PD-L1 (≥50%), whereas responses were comparable in patients with low/medium TMB irrespective of PD-L1 expression.[5] Surprisingly, given the strength of the tumor response and PFS signals, high TMB failed to differentiate between nivolumab and chemotherapy in terms of OS, although in absolute terms, median OS (mOS) was greater in patients with high TMB in both the nivolumab and chemotherapy arms.

The Phase III CheckMate-227 trial (nivolumab plus ipilimumab vs. chemotherapy) showed similar findings: Whilst there was no significant difference in OS benefit between TMB >10 mut/Mb or <10 mut/Mb, there was a trend toward greater OS in subjects with high TMB (HR 0.68, mOS 23 months) compared to low TMB (HR 0.75, mOS 16.2 months).[8]

These findings were further validated in the NEPTUNE trial, where TMB failed to differentiate a survival benefit with durvalumab plus tremelimumab vs. chemotherapy.[9]

Where does this leave TMB as a biomarker for response?

TMB has yet to receive regulatory approval for guiding therapeutic decision making. Despite the increasing number of studies evaluating the potential clinical relevance of TMB as a predictive biomarker for immunotherapy response, its use in the clinical setting is currently limited by the absence of standard methods of quantification and the lack of a robust and universal cutoff to identify immunotherapy responders. Multiple trials that have evaluated TMB (KEYNOTE-12, KEYNOTE-028, IMvigor210, IMvigor211, CheckMate-012 and CheckMate-568) across different tumor types to date have used different cutoff values to define high TMB, adding to the confusion.

As a biomarker that affects treatment decisions, it is essential to unify TMB assessment approaches to allow for reliable, comparable results across studies. When implementing TMB measurement assays, it is important to consider factors that may impact the method workflow, the results of the assay and the interpretation of the data. Such factors include biopsy sample type, sample quality and quantity, genome coverage, sequencing platform and, ultimately, defining the threshold that determines high TMB.


  1. Paz-Ares, Luis. “Pembrolizumab (pembro) plus platinum-based chemotherapy (chemo) for metastatic NSCLC: tissue TMB (tTMB) and outcomes in KEYNOTE-021, 189, and 407.” ESMO Oncology-PRO. (accessed October 15, 2019).
  2. Herbst, Roy S. “Association between tissue TMB (tTMB) and clinical outcomes with pembrolizumab monotherapy (pembro) in PD-L1-positive advanced NSCLC in the KEYNOTE-010 and -042 trials.” ESMO Oncology-PRO. (accessed October 15, 2019).
  3. Socinski, Mark. “Final efficacy results from B-F1RST, a prospective Phase II trial evaluating blood-based tumour mutational burden (bTMB) as a predictive biomarker for atezolizumab (atezo) in 1L NSCLC.” ESMO Oncology-PRO. (accessed October 15, 2019).
  4. Reck, Martin, et al. “Updated Analysis of KEYNOTE-024: Pembrolizumab Versus Platinum-Based Chemotherapy for Advanced Non–Small-Cell Lung Cancer With PD-L1 Tumor Proportion Score of 50% or Greater.” Journal of Clinical Oncology, Vol. 37 (2019): 537-546.
  5. Carbone, David P., et al. “First-Line Nivolumab in Stage IV or Recurrent Non–Small-Cell Lung Cancer.” New England Journal of Medicine, Vol. 376 (2017): 2415-2426.
  6. Mok, Tony S. K., et al. “Pembrolizumab versus chemotherapy for previously untreated, PD-L1-expressing, locally advanced or metastatic non-small-cell lung cancer (KEYNOTE-042): a randomised, open-label, controlled, phase 3 trial.” Lancet, Vol. 393 (2019): 1819-1830.
  7. Spigel, David. “IMpower110: Interim overall survival (OS) analysis of a phase III study of atezolizumab (atezo) vs platinum-based chemotherapy (chemo) as first-line therapy.” ESMO Oncology-PRO. (accessed October 15, 2019).
  8. Bristol-Myers Squibb. “Bristol-Myers Squibb Provides Update on the Ongoing Regulatory Review of Opdivo Plus Low-Dose Yervoy in First-Line Lung Cancer Patients with Tumor Mutational Burden ≥10 mut/Mb.” Press release, October 19, 2018. (accessed October 20, 2019).
  9. Astor, Lisa. “OS Results Fall Flat With Durvalumab/Tremelimumab Combo in NEPTUNE Trial.” Targeted Oncology. (accessed November 15, 2019).

Back to Resources