Edwin Basak

GZMB are associated with PFS, OS, or BOR, in patients with NSCLC treated with nivolumab.

Patients with low baseline levels of granzyme B had a shorter PFS (HR: 1.96; 95% CI: 1.12 – 3.43; p = 0.018) compared to patients with high baseline levels of granzyme B. Moreover, patients with low levels of baseline granzyme B had shorter OS (HR: 2.08; 95% CI: 1.12 – 3.87; p = 0.021) compared to patients with high levels of granzyme B. Of all investigated SNPs, patients with homozygous or heterozygous variants of GZMB rs8192917 had significantly shorter PFS (HR: 1.38; 95% CI: 1.02 – 1.87; p = 0.036) compared to wildtype patients. Moreover, the same group of patients had worse BOR (OR: 1.60; 95% CI: 1.01 – 2.52; p = 0.044). None of the SNPs were associated with OS. Of note, patients with homozygous and heterozygous variants of the rs8192917 SNP had significant lower baseline serum levels of granzyme B (mean difference 8.6 pg/mL; 95% CI: 1.77 – 15.38; p = 0.015). Before mentioned findings of chapter 8 might indicate a potential role for granzyme B in assessing which patients are less likely to respond to anti-PD-1 therapy. Especially, since both serum levels of granzyme B and genetic variations in GZMB are associated with treatment outcomes.

Within the MULTOMAB-trial, we had the ability to stain peripheral blood samples with multiplex flow cytometry, providing the ability to investigate immune cell subsets in different cohorts of patients. Relevance of such analysis might follow from results of tumor biopsies, which showed that CD8 T cell density predict response to anti-PD-1 therapy in patients with advanced melanoma. However, tumor biopsies are quite invasive to obtain. Also, in ipilimumab treated patients with melanoma, it has been shown that the absolute number of lymphocytes in peripheral blood samples correlates with clinical outcome. At last, Ki67 expression in a subset of PD-1 + + CD8 T cells has been shown to portray effector T cell invigoration in patients with advanced melanoma and NSCLC treated with anti-PD-(L)1 therapy. 59-61 Following abovementioned results, in chapter 9, this thesis describes 18 immune cell subsets and assessed the expression of 28 T cell markers in patients with NSCLC cancer who were treated with 2nd line nivolumab.

This study showed that in patients with NSCLC treated with nivolumab and having a PR a higher number of CD8 T cells was found. Furthermore, those patients had a phenotype which corresponds to a late differentiation at baseline. This differentiation was accompanied by a higher frequency of PD-1 and TIMP, together with a complete loss of co-stimulatory receptors. The results in chapter 9 were one of the first using peripheral blood to describe peripheral immune markers to be able to identify patients with NSCLC showing a response to nivolumab therapy.

Thyroid toxicity is one of the most occurring AE during anti-PD-1 therapy, occurring in around 5-15% of the patients. 62-64 Earlier studies had suggested that thyroid toxicity during anti-PD-1 therapy was associated with improved OS, however those studies included only small sample sizes and showed conflicting results regarding PFS. 65,66 Furthermore, the presence of anti-thyroid antibodies (ATAbs), before anti-PD-1 therapy has been related to the development of thyroid toxicity during anti-PD-1 therapy, 67,68 while it was not correlated with improved OS. In chapter 10 of this thesis, the hypothesis that the occurrence of thyroid toxicity during anti-PD-1 therapy and that the presence of ATAbs may predict anti-PD-1 treatment outcomes is investigated, among patients with NSCLC, metastatic melanoma, and RCC.

The study described in chapter 10 showed that patients with acquired overt thyroid dysfunction had longer OS (HR: 0.18; 95% CI: 0.04 - 0.76; p = 0.020) and PFS (HR: 0.39; 95% CI: 0.15 – 0.998; p = 0.050) compared to patients without thyroid dysfunction. Moreover, ATAb levels at baseline did not correlate with either OS or PFS. However, patients with higher ATAb levels during anti-PD-1 treatment had improved OS (HR: 0.39; 95% CI: 0.21 – 0.72; p = 0.003) and PFS (HR: 0.52; 95% CI: 0.33 – 0.81; p = 0.004). While acquired overt thyroid dysfunction and higher ATAb levels during anti-PD-1 treatment correlate with OS and PFS, baseline ATAb levels did not correlate with treatment outcomes, limiting the use of ATAb levels at baseline to estimate anti-PD-1 treatment outcomes.

CONCLUSIONS AND FUTURE PERSPECTIVES

Typical of treatment with ICIs is the phenomenon of long lasting, durable responses in a proportion of patients, while many other patients do not benefit from therapy at all. Furthermore, at time of introduction of ICIs, a lot was unknown about if patient and clinical characteristics, biomarkers, or drug pharmacokinetics could predict which patients were more likely to benefit from treatment with ICIs or were more prone to experience side effects. This thesis described one of the first efforts to answer those questions, using mostly an hypothesis generating approach by using the MULTOMAB-study, a multicenter multi histology observational cohort study. With the registration of the first anti-PD-1 antibodies, this study was initiated. The MULTOMAB-study provided the ability to set up a large biobank, storing clinical data and sequential patient serum and peripheral blood mononuclear cells (PBMCs). This allowed to investigate the correlation between treatment outcomes and anti-PD-1 pharmacokinetics, biomarkers used in routine clinical care, and immunological markers. While in the early days of anti-PD-1 treatment hypothesis generating research may have narrowed down which areas of interest to focus on, it is necessary that future studies should focus on the results of those earlier hypothesis generating studies. Furthermore, prospective trials will allow for uniform and more carefully documented study outcomes.

Anti-PD-1 therapy is increasingly used in combination with other therapies (e.g. anti-CTLA-4 therapy or chemotherapy), 70,71 a trend that will continue in coming years. It is plausible that markers correlating with minimal differences in treatment outcomes, being response or adverse events, will not dramatically influence clinical decision making in patients without other treatment options. However, characteristics related to the occurrence of adverse events might impact the decision to administer more invasive combination regimens or more tolerable monotherapy. As such, future prediction studies could focus on this distinction.

Many studies have looked at individual biomarkers and clinical characteristics to assess factors associated with AEs or response rates. However, many conflicting results do exist among those, mostly retrospective, studies. For example, considering age and sex, both positive and negative significant associations were found. Two other reviews found great heterogeneity between individual study outcomes, while only a few findings were validated in an independent cohort. 72,73 Adding to this, no uniform outcome measures are used. For example, studies investigating associations between SNPs and AEs, all use different outcome measures (i.e. all grade AEs vs. no AEs, or grade ≥3 AEs vs no AEs, etc.) or they use different definitions per type of AE. 39,40,42,43 To overcome this matter, standardised and pre-defined outcomes should be used in coming studies. Also, prospective studies should be performed to reliably score and collect clinical outcomes such as grade of AEs.

Using knowledge over the past years from the MULTOMAB-study and other performed studies, prospective studies have been started, such as the NutriCim and NIVOPTIMIZE trials. The NutriCim study aims to assess whether a nutritional intervention might impact pembrolizumab clearance. An important question given the results of chapter 3 and chapter 4, and more specifically given the findings in Hurkmans et al. 74 Which found an inverse clearance-OS relationship for NSCLC patients treated with pembrolizumab. This study might elucidate the true relationships between cachexia, response, and nivolumab pharmacokinetics. The NIVOPTIMIZE study will compare steady-state levels of a reduced dose with nivolumab serum levels that are considered effective. Aiming to pave the way for larger studies in which reduced nivolumab dosages will be administered, leading to reduced treatment costs. Since costs of expensive treatment options have almost doubled in the period from 2012 to 2019, 75 studies to introduce those treatment options more efficiently in selected subgroups should remain in the future.

The protocol of the MULTOMAB-study allows the inclusion of new monoclonal antibodies and tumor types. While the availability of a large set of data and samples allows research to rare conditions, as shown in chapter 7, translation to the clinic remains a big issue, since the retrospective nature and hypothesis generating concept of the study makes uniform outcome measures difficult to assess. However, as demonstrated in chapters 3 and 4, having a selected number of patients included from each treatment type and tumor type might lead to interesting hypothesis. However, such retrospective analyses should be followed by larger prospective trials, with uniform outcome measures and being adequately powered. Leading to useful and general conclusions and implications for daily clinical practice. Therefore, there remains a place for hypothesis generating studies with a large data- and sampleset. However, a clear pre-existent hypothesis should be in place, while collecting just sufficient data and patient material to answer or test those specific hypotheses. New law regulations, such as the “Wet zeggenschap lichaamsmateriaal” are under debate, 76,77 and might further regulate the broad use of patient data and samples without a clear pre-defined hypothesis. By collecting data and patient material more specifically related to a pre-defined hypothesis, studies such as the MULTOMAB-study could be implemented more time and cost efficiently, sparing resources for larger prospective trials.