Predictive biomarkers are critical for patient selection in immune checkpoint inhibitor (ICI) therapy. However, the term “predictive” are often confused with “prognostic,” leading to inappropriate treatment choice in clinical practice and research.

We read with great interest the article by Aggarwal and colleagues reporting that blood-based tumor mutational burden (bTMB) plus putative negative predictors (e.g., STK11 and KEAP1 mutations) could significantly predict both progression-free survival (PFS) and overall survival (OS) benefits of pembrolizumab-based therapy in non–small cell lung cancer (NSCLC; ref. 1).

Retrospective analyses in the POPLAR/OAK and NCC cohorts and prospective analysis in the B-F1RST trial both demonstrated that unadjusted bTMB was associated with ORR/PFS, but not OS, upon immunotherapy (2, 3). Inconsistent with previous reports, Aggarwal and colleagues found that bTMB alone was nonsignificantly associated with the OS benefit of pembrolizumab (P = 0.061; ref. 1). Thus, the authors introduced putative negative predictors, some of which were more prognostic than predictive, for example, STK11 and KEAP1 mutations (4, 5). Because of their prognostic effects, the authors came to the conclusion that bTMB plus negative predictors can precisely identify patients with OS benefits from pembrolizumab-based therapy in NSCLC.

Taking STK11 mutation for example, we calculated pooled estimates from six ICI-treated cohorts with patient-level data of 807 EGFR/ALKWT LUAD cases, demonstrating that STK11 mutation was associated with significantly shorter PFS (HR = 1.54; 95% CI 1.17–2.03, P = 0.002) and OS (HR = 1.57; 95% CI 1.16–2.11, P = 0.003), but nonsignificantly lower ORR (RR = 0.80; 95% CI 0.43–1.47). Likewise, in STK11-mutated patients receiving docetaxel, worse ORR and PFS were observed with borderline significance while OS was significantly shorter (HR = 1.82; 95% CI 1.18–2.80, P = 0.006). Particularly, in the POPLAR/OAK trials comparing atezolizumab with docetaxel, STK11 aberration and treatment group had independent effects on OS, as shown by a nonsignificant interaction between these two variables (HR = 1.10; 95% CI 0.60–2.01; P = 0.766), indicating that STK11 mutation is prognostic rather than predictive. Furthermore, in the TCGA database, STK11 mutation was associated with poorer prognosis (HR = 1.57; 95% CI, 1.07–2.31, P = 0.021), with an HR similar to those seen in other ICI-treated cohorts.

Predictive biomarkers should predict therapeutic efficacy without interference from their prognostic effects. In the POPLAR/OAK studies, atezolizumab conferred similar OS benefits over docetaxel in both STK11-mutated (HR = 0.66) and STK11-wildtype (HR = 0.59) patients with LUAD, indicating that STK11 mutation should not serve as a basis for treatment choice between ICIs and chemotherapy. Collectively, we recommend prognostic biomarkers being excluded from efficacy prediction in studies without a control arm to avoid misleading conclusions.

See the Response, p. 3492

No potential conflicts of interest were disclosed.

This work was supported by National Natural Science Foundation of China (81974361, to X. Li) and CAMS Innovation Fund for Medical Sciences (2018-I2M-1-002, to L. Li). We sincerely thank Chan Gao for English language editing.

1.
Aggarwal
C
,
Thompson
JC
,
Chien
AL
,
Quinn
KJ
,
Hwang
WT
,
Black
TA
, et al
Baseline plasma tumor mutation burden predicts response to pembrolizumab-based therapy in patients with metastatic non-small cell lung cancer
.
Clin Cancer Res
2020
;
26
:
3492
.
2.
Wang
Z
,
Duan
J
,
Wang
G
,
Zhao
J
,
Xu
J
,
Han
J
, et al
Allele frequency-adjusted blood-based tumor mutational burden as a predictor of overall survival for patients with NSCLC treated with PD-(L)1 inhibitors
.
J Thorac Oncol
2020
;
15
:
556
67
.
3.
Kim
ES
,
Velcheti
V
,
Mekhail
T
,
Leal
TA
,
Dowell
JE
,
Tsai
ML
, et al
Primary 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 non-small cell lung cancer (NSCLC)
.
Ann Oncol
2018
;
29
:
viii744
.
4.
Goeman
F
,
De Nicola
F
,
Scalera
S
,
Sperati
F
,
Gallo
E
,
Ciuffreda
L
, et al
Mutations in the KEAP1-NFE2L2 pathway define a molecular subset of rapidly progressing lung adenocarcinoma
.
J Thorac Oncol
2019
;
14
:
1924
34
.
5.
Kadara
H
,
Choi
M
,
Zhang
J
,
Parra
ER
,
Rodriguez-Canales
J
,
Gaffney
SG
, et al
Whole-exome sequencing and immune profiling of early-stage lung adenocarcinoma with fully annotated clinical follow-up
.
Ann Oncol
2017
;
28
:
75
82
.