The European Society for Medical Oncology (ESMO) held its annual immune-oncology symposium in Lausanne on Friday 20th and Saturday 21st November 2015, providing a great opportunity for anyone interested to find out about all aspects of immune therapies and latest developments.
As would be expected, together with reviews on the activity of checkpoint inhibitor antibodies (anti-PD1, -PD-L1 and -CTLA4) and various cell and vaccines therapies, discussions on personalised medicine and predictive biomarkers were also at the forefront. At present, checkpoint inhibition appears to be efficacious in, roughly, one third of the patients treated with such agents. For many reasons it is therefore essential to have the tools to able to select or stratify the right patients for immunotherapy, at the right time and at the right dose.
For oncology physicians and scientists, especially those involved in the development of targeted therapies, some of the issues with immunotherapy predictive biomarkers sound very familiar. At the start of the meeting Professor Gordon Freeman (Dana-Farber Cancer Institute) reviewed a collection of clinical studies involving either anti-PD1 antibodies (nivolumab, pembrolizumab) or anti-PD-L1 antibodies (atezolizumab, MEDI4736) in various tumour types and showing that PD-L1 positivity, as assessed by immunohistochemistry (IHC), increases the likelihood of response to PD-1/ PD-L1 blockade. He cautioned though that the particular areas of the biopsy used to assess PD-L1 positivity are critical determinants of the final outcome. He referred to the study by Callea et al., 2015 which demonstrated that there is 21% discordance in PD-L1 expression between primary and metastatic renal cell carcinoma and furthermore PD-L1 positivity was almost exclusively detected in high nuclear grade areas.
In their presentations, Dr. David Feltquate (Bristol-Myers Squibb) and Professor Solange Peters (Centre Hospitalier Universitaire Vaudois) discussed how different IHC tests are being applied by different companies in clinical trials. Interpretation of the extent (strong versus weak) and degree (≥1%, ≥5% or ≥50%) of expression, and the cut-offs selected, vary. Professor Peters referred to PD-L1 positivity as a “flexible concept”. Nivolumab (Bristol-Myers Squibb) clinical trials have employed the Dako 28-8 antibody to score tumour cell membranes in archival or fresh tissues; pembrolizumab (Merck) is using the Dako 22C3 antibody to score tumour cell membranes in fresh tissues, atezolizumab (Roche/ Genentech) is using the Ventane SP142 antibody to score tumour cell membranes or infiltrating immune cells in archival or fresh tissues and durvalumab (AstraZeneca) is using the Ventane SP263 antibody to score tumour cell membranes archival or fresh tissues.
Slightly adding to the confusion is the fact that in October 2015 the US Food and Drug Administration (FDA) approved Dako 28-8 PharmDX as a complementary diagnostic and DAKO22C3 as a companion diagnostic. The latter is a diagnostic typically associated with a particular drug and is included in its approval label, whereas the former is more broadly associated with a class rather than a specific drug and is not restricted by labelling; each category having particular ensuing regulatory and economic consequences (Milne et al., 2015).
Dr. Feltquate commented that even though PD-L1 correlates with tumour response and appears to correlate with overall survival in some tumour types (e.g. in NSCLC and bladder cancer), an appropriately designed phase III trial with a biomarker-based primary endpoint in order to fully validate the clinical utility of the biomarker is yet to be conducted. Hence, PD-L1 expression, as assessed currently, may have value for regulatory and payer access in some tumour types but is not useful for clinical decision making i.e. better biomarkers are needed.
Dr. Ramy Ibrahim (AstraZeneca) noted that an initiative being led by the FDA-ASCO-AACR and in collaboration with the pharmaceutical industry is currently underway which is aiming to harmonise the PD-L1 assay and to generate a “PD-L1 blueprint”.
PD-L1 expression differs between tumour and immune cells not only in terms of expression but also in terms of biology. This is part of the bigger interplay between the “tumour cell army” and the “T-cell army” within the tumour microenvironment (Groopman 2012). Referring to this concept, Dr. William Grossman (Genentech/Roche) proposed a personalised cancer immunotherapy algorithm that prescribes anti-PD-1 or anti-PD-L1 therapy in combination with other modalities (targeted therapies, chemo- or radiotherapy or other immunotherapies) which also takes into account the tumour microenvironment and is based on the observation that inflamed tumours may respond better to immune checkpoint inhibition than non-inflamed tumours (Woo et al., 2015).
Besides PD-L1, another possible biomarker that received a lot of attention during the symposium was the mutation load of a tumour and more specifically in relation to ‘neo-antigens’ i.e. epitopes that arise as a consequence of tumour-specific mutations. The main impact of such efforts may not be in development of novel biomarkers as analysis of each patients tumour mutatome is at present time and costly prohibiting. Importantly, however, it is being exploited in guiding towards personalised or precision immunotherapy and even individualised (immune)therapy for cancer.
Professor Ton Schumacher (Netherlands Cancer Institute) presented an overview of data generated by employing immune response detection (neo-antigen specific CD8 T-cell responses) in a high throughput manner and indicating that mutational load correlates with clinical outcome to anti-PD-1 in NSCLC. Studying the biology of such neo-antigenic responses is enabling the generation of CAR-T or synthetic vaccines that are now in the clinic (Schumacher and Schreiber, 2015). Such an individualised approach was discussed by Professors Ugur Sahin (BioNTech AG) and Pierre-Yves Dietrich (University of Geneva), both members of the GAPVAC consortium that is currently testing in the clinic (Phase II studies) synthetic RNA-based and pharmacologically optimised vaccines for adoptive cell therapy. Such vaccines, derived from mutated peptides exclusively presented on human leukocyte antigen (HLA) receptors on tumour cells are manufactured on an individual patient-specific manner.
It seems that beyond immune checkpoint inhibition and together with improvements in high throughput R&D and improved GMP manufacturing of cellular and molecularly-based technologies, in industry and academia, immune-oncology is gradually progressing personalised and precision to truly individualised medicine. This can only be good for cancer patients!
Callea M, Albiges L, Gupta M, Cheng SC, Genega EM, Fay AP, Song J, Carvo I, Bhatt RS, Atkins MB, Hodi FS, Choueiri TK, McDermott DF, Freeman GJ, Signoretti S. Differential Expression of PD-L1 between Primary and Metastatic Sites in Clear-Cell Renal Cell Carcinoma. Cancer Immunol Res. 2015 Oct;3(10):1158-64.
Groopman J. The T-cell army. Can the body’s immune response help treat cancer? The New Yorker, April 23.
Milne CP, Bryan C, Garafalo S, McKiernan M. Complementary versus companion diagnostics: apples and oranges? 1. Biomark Med. 2015;9(1):25-34.
Schumacher TN, Schreiber RD. Neoantigens in cancer immunotherapy. Science. 2015 Apr 3;348(6230):69-74.
Woo SR, Corrales L, Gajewski TF. The STING pathway and the T cell-inflamed tumor microenvironment. Trends Immunol. 2015 Apr;36(4):250-6.
Reference photo: Illustration by Victo Ngai – Taken from Groopman J. The New Yorker, April 23.