In recent years, immunotherapy has changed the treatment landscapes of lung cancer and prolonged survival of patients. However, PD-1 or PD-L1 monoclonal antibodies have limited efficacy, about 20% of NSCLC patients could get tumor response to immunotherapy.28 BsAbs can simultaneously target two different checkpoints, which may improve efficacy of immunotherapy and become a promising innovative therapy for cancer patients.29 KN046 is a BsAb targeting PD-L1 and CTLA4, demonstrating efficacy in 2nd line or beyond treatment of advanced NSCLC patients in the phase II trial.13 However, its predictive biomarkers are largely unknown. This study demostrated stroma signature based on protein expression is associated with better efficacy of the compound.
Combination of PD-1/PD-L1 and CTLA4 antibodies is synergistic in antitumor effect. The efficacy of this dual immune combination strategy was firstly demonstrated in melanoma patients.30 In NSCLC, CheckMate227 study suggested that patients could benefit from dual immunotherapy.31 In CheckMate9LA, overall survival (OS) of nivolumab and ipilimumab and 2 cycles chemotherapy was higher than chemotherapy alone. But the combination of nivolimab and ipilimumab produces significant toxicity, which is mainly due to ipilimumab.32 Several bsAbs against PD-1/PD-L1 and CTLA4 with the purpose to decrease toxicity of the combination while keep efficacy are being tested.13, 33, 34 Just like previous PD-L1 or PD-1 inhibitors, not all patients benefit from bsAbs. Up to now, there was no good biomarkers to screen out effective population of bsAbs.
In the process of exploring tumor biomarkers of PD-1/PD-L1 monoclonal antibodies in the past, several biomarkers related to immunotherapy efficacy were found, but all had their limitations. PD-L1 was the most widely used predictor so far. Many studies have noted that patients with higher PD-L1 expression could benefit from immunotherapy, they can achieve longer PFS or OS.35–38 Another common biomarker is TMB. Several studies revealed that patients with higher TMB had higher objective response rate (ORR) and longer PFS.39–41 TMB is closely related to DNA repair defects. Generally, patients with high TMB often have high DNA mismatch repair (dMMR) and high microsatellite instability (MSI-H). A large number of studies have confirmed that patients with advanced solid tumors of MSI-H/dMMR are more likely to benefit from immunotherapy, regardless of cancer type.42, 43 Tumor infiltrating lymphocytes (TILs), as a kind of highly heterogeneous immune cells, has been shown to be a favorable prognostic factor for many solid tumors, including lung cancer.44 Precious studies suggested that intense TILs were independent prognostic variable for PFS and OS.45–47 But the lack of globally consistent criteria and cutpoints may reduce their predictive value. Interferon gamma (IFNγ) related signaling pathways play an important role in exerting anti-tumor immune effects. Patients with high baseline IFN-γ mRNA expression may have longer PFS and OS.48, 49 However, IFN-related studies mainly focused on melanoma, and data on NSCLC are limited. Then several studies also focused on the relationship between the combination of PD-L1 + TMB and the efficacy of immunotherapy. In CheckMate 568, patients with high TMB + negative PD-L1 showed higher ORR than those with high TMB + positive PD-L1.50 On the contrary, CheckMate 026 suggested that in PD-L1 positive + high TMB subgroup, the response rate was up to 75%, which was higher than only one factors.51 Therefore, whether the combination of biomarkers can predict the curative effect remains to be further explored. We need a more advanced platform to access more information about tumor tissue.
DSP technology solved the problem of spatial information detection. The main advantages of DSP was the high reuse capability for FFPE samples, the low operating time, and the non-destructive direct process to define biomarkers in a discrete ROI.20, 21, 52 In the field of immunotherapy companion diagnostics, Gupta S. et al. and Zugazagoitia J. et al. have confirmed that DSP seems to have quantitative potential compared to IHC, and the technology has the capability to do concomitant diagnostic tests for immunotherapy.53, 54 DSP was also used in finding biomarkers. It was shown fisrstly to successfully identify 20 biomarkers in melanoma patients, where the expression of PD-L1 in CD68-positive cells rather than tumor cells was an important factor in determining PFS, OS, and treatment response.55 DSP seems to have the potential to become an accurate technique for determining therapy prognosis after immunotherapy, so we used it to detect prognostic biomarkers from bsAbs.
The composition of tumor microenvironment (TME) plays an important role in inhibiting or enhancing immune response. Tumor stroma is an important part of tumor microenvironment, which plays an important roal of tumor genesis, development and metastasis.56 The precision of tumor immunotherapy and the exploration of biomarkers should not only focus on tumor cells, but also the stroma information around them and the spatial information of each component should not be ignored. This study firstly validated the immune components and protein expressions in different tumor areas and stroma areas were significantly different using DSP, which also laid a foundation for us to determine more accurate biomarkers. In provious study about NSCLC patients treated with dual immunotherapy combination, CheckMate227 suggested that nivolumab plus ipilimumab improved PFS in patients with high TMB (≥ 10), and prolonged OS in patients with positive PD-L1 expression (≥ 1%).31, 57 When researches made a combination of these two biomarkers, patients with high TMB + positive PD-L1 achieved longer PFS, but did not increase PFS when compared to the high TMB + negative PD-L1 subgroup.31 The combination of PD-L1 and TMB may not make sense. But in our study, we identified a predictive signature of Tim-3, PD-L1, CD11c, and B2M in stroma areas with higher prediction score than tumor cells. Its score exceeded traditional biomarkers PD-L1 and TMB. This suggests that bsAbs immunotherapy may alter the TME and induce greater changes in protein expression of stroma regions. This combination of spatial different proteins may ameliorate some drawbacks of traditional biomarkers. It also provide new method and idea for the exploration of efficacy predictive biomarkers of other bsAb drugs.
Despite our foundings above, this study also had its limitations. When we verified the relative genes and signatures in common database, we were unable to obtain positive results. It may be caused by the following reasons. First, The validation sets focused on multiple cancer types with immune monotherapy, which is different from our target population and treatment regimen. Second, we could only got data at RNA level, it was different from the protein level. Besides, because of the novelty of KN046, relative data can not be obtained from public database. Furthermore, the sample size was small and involved one single center, and due to the amount of collected tissue, we only carried out DSP for protein detection. So our signature of bsAb-KN046 needs to be further validated in subsequent prospective studies and clinical trials involving multiple centers. And in the future, we plan to add other techniques such as single-cell sequencing and spatial transcriptome sequencing to further verify our findings.
In conclusion, our study suggest that there are differences in the expression of immune-related protein information in different spatial regions of tumor tissue, and the signature in stroma region with Tim-3, PD-L1, CD11c, and B2M can better predict the response of a bsAb drug KN046.