PD-L1 expression and prediction of response to immune modulators in non-small cell lung cancer; reasons for its fragility and strategies to reduce it



Haragan, Alexander
(2021) PD-L1 expression and prediction of response to immune modulators in non-small cell lung cancer; reasons for its fragility and strategies to reduce it. PhD thesis, University of Liverpool.

[img] Text
201313407_May2021.pdf - Unspecified

Download (25MB) | Preview

Abstract

Abstract Introduction Anti-PD-1/PD-L1 immunomodulatory (IM) therapy has revolutionised the treatment of non-small cell lung cancer (NSCLC). The only ‘biomarker’ currently-validated for predicting response of these tumours to IM therapy is the extent of PD-L1 expression as detected by immunohistochemistry (IHC). Despite the overall success of this therapy in patients with NSCLC, PD-L1 expression is an imperfect predictor, some patients with tumours displaying low expression responding strongly, and some with high expression not at all. The thesis considers why PD-L1 expression is an imperfect predictor and how it might be improved. Methods The research described in the first part of this thesis considered the impact of pre-analytical conditions on PD-L1 expression. This examined not only the effect of how tumours are sampled, but the influence of specimen processing and fixation and conditions of storage, the latter employing a novel tissue ageing acceleration chamber and mass-spectrometry. The second part describes examination of heterogeneity of expression in a series of resected NSCLCs in which the primary tumour was accompanied by nodal metastases. Biological and artefactual heterogeneity within and between tumour deposits was assessed at different scales using a novel ‘squares method’ and ‘digital sampling’. The third part describes assessment of the tumour immune environment (TME), specifically interrogation of immune cell populations, employing a combination of techniques including traditional IHC, multiplex IHC, multiplex immunofluorescence and image analysis. The fourth and final part of the work involved an assessment of digital pathology and image analysis with integrated machine-learning algorithms as a tool to improve accuracy and consistency in assessing PD-L1 expression. Results PD-L1 expression is consistent across different types of specimen; loss of its immunogenicity can be reduced by storage in cold and dry conditions, particularly when combined with a desiccant. Approximately 20-25% of resected NSCLCs demonstrated tumoural heterogeneity such that sampling from different sites might produce clinically-relevant differences in PD-L1 expression. This can be minimised, but not reduced entirely, by generous sampling. The TME of NSCLCs can be differentiated by assessing different immune cell populations, but only in specimens containing sufficient tissue and routine, small, diagnostic specimens will prove difficult to analyse in this way. Image analysis and algorithms are potentially powerful tools that can reduce intra- and inter-observer consistency when assessing PD-L1 expression, but require learning and experience for their effective use. Discussion The research described in this thesis confirms that assessment of PD-L1 expression by IHC is a powerful, but imperfect biomarker, and indicates also that its utility can be improved. Accuracy and consistency in its interpretation can be increased by optimising pre-analytical conditions. Tumour heterogeneity is a more complex problem; whilst availability of multiple, generous, good quality samples improves accuracy, the confounding effect of this fundamental fact of the biology of PD-L1 expression cannot be removed entirely. Techniques to interrogate the TME yield powerful data but, at present, most are too expensive, too complicated and require too much tissue to be useful in the routine clinical setting. Image analysis, machine learning and algorithms are becoming established techniques and are clearly of value, but possibly largely in improving confidence in and consistency of interpretation.

Item Type: Thesis (PhD)
Divisions: Faculty of Health and Life Sciences > Institute of Systems, Molecular and Integrative Biology
Depositing User: Symplectic Admin
Date Deposited: 27 Jul 2021 13:08
Last Modified: 22 Nov 2021 12:27
DOI: 10.17638/03125235
Supervisors:
URI: https://livrepository.liverpool.ac.uk/id/eprint/3125235