A Novel Class of On-Treatment Cancer Immunotherapy Biomarker: Trough Levels of Antibody Therapeutics in Peripheral Blood

ABSTRACT

While immune checkpoint blockade has revolutionized cancer treatment, unfortunately most patients do not benefit from this treatment. Many pharmacodynamic (PD) studies have revealed essential requirements for successful cancer immunotherapy that may provide insight into how we can improve these agents. Despite enormous efforts focused on interrogating the immune system using different biospecimens (e.g. blood, primary tumor, metastatic tumor, microbiome samples), a variety of technologies (e.g. flow cytometry, bulk and single-cell RNA-sequencing, immunohistochemistry), and wide-ranging disciplines (e.g. pathology, genomics, bioinformatics, immunology, cancer biology, metabolomics, bacteriology), discovery of consistent biomarkers of response have remained elusive. Pharmacokinetics (PK) studies, however, not only provide critical information regarding safe dosing but may also reveal useful biomarkers. For example, recent studies found that trough levels of therapeutic monoclonal antibodies (mAbs) or clearance (CL) of them were associated with clinical outcome, which suggests that trough levels of mAbs may represent a new class of on-treatment cancer immunotherapy biomarker. In this review, we summarize the potential utility of trough levels of mAbs, the mechanism of varying PK, consideration for therapeutic drug monitoring, and assay attributes that will facilitate wider utilization of PK information in conjunction with PD assessments.

KEYWORDS:

• Antibody therapeutics
• biomarkers
• cancer immunotherapy
• clearance
• immune checkpoint blockade
• inflammation
• pharmacodynamics
• pharmacokinetics
• trough levels

Introduction

Cancer is a genetic disease caused by accumulation of somatic mutations in oncogenes and tumor-suppressor genes despite the orchestrated repair of DNA replication errors. These mutations can generate a source of neoantigens suitable for recognition by T cells (Schumacher and Schreiber 2015). Based on this notion, efforts to augment host anti-tumor immunity as a therapeutic strategy have been fruitful. For example, high-dose interleukin-2 (IL-2) and adoptive T cell therapy can produce significant clinical responses in a subset of patients with melanoma and renal cell carcinoma (Jiang et al. 2016; Rosenberg and Dudley 2009). Furthermore, a breakthrough came from the discovery that tumor-specific T cells are negatively regulated by immune checkpoint molecules such as CTLA-4 and PD-1. This has resulted in the groundbreaking success of immune checkpoint blockade (ICB) with anti-CTLA-4 or anti-PD-1/PD-L1 antibodies (Ledford et al. 2018), which enable active anti-tumor immune cells. Anti-CTLA-4 was approved for advanced melanoma in 2011, however its use is limited due to the high frequency of severe immune adverse events. Anti-PD-1 was first approved for advanced melanoma in 2014 and, since then, an additional six anti-PD-1/PD-L1 antibodies have been approved by the US FDA for multiple indications, including the pan-cancer indication of microsatellite instability-high (MSI-H) or deficient mismatch repair (dMMR) tumors. While these successes have represented a meaningful advance in cancer therapy, only a minority of patients respond to anti-PD-1/PD-L1 antibodies, likely due to the reliance of anti-PD-1/PD-L1 antibodies on the presence of pre-existing anti-cancer immunity (Robert 2020).

PD studies using flow cytometry, bulk RNA-seq analysis, or immunohistochemistry of tumor tissue revealed that better clinical outcome of ICB treatment is associated with T cell activation and proliferation, PD-L1 checkpoint expression in the tissue, and higher genomic mutation burden (Figure 1). For anti-CTLA-4, higher absolute lymphocyte counts in peripheral blood (Ku et al. 2010), increased inducible T-cell costimulator (ICOS, CD278) expression in CD4⁺ T cells during treatment (Ng Tang et al. 2013), higher tumor mutational burden (Snyder et al. 2014), and an increased post-treatment interferon gamma (IFN-γ) signature in the tumor were associated with longer overall survival (OS) (Gao et al. 2016). We also found that pretreatment serum levels of the chemokine CXCL11 and soluble MHC class I polypeptide–related chain A (sMICA) were negatively associated with worse outcomes in patients with advanced melanoma upon receiving ipilimumab (Koguchi et al. 2015). For anti-PD-1, high CD8⁺ T-cell density in the tumor (Tumeh et al. 2014), high PD-L1 expression in the tumor (Topalian et al. 2012), microsatellite instability high tumor (Le et al. 2015), baseline IFN-γ-related gene signatures (Ayers et al. 2017), and high CD8+ T cell proliferation in the periphery in conjunction with lower tumor burden correlated with longer OS (Huang et al. 2017). Recent single-cell RNA-sequencing (scRNA-seq) analysis of tumor biopsies also revealed an increased level of precursor exhausted T cells in responsive tumors post-treatment that was accompanied by low expression of inhibitory molecules and high expression of granzyme K (Liu et al. 2022). Together, these PD studies and others have provided critical information to understand the mechanism of action for ICBs, which can also be used as biomarkers.

Figure 1. PK biomarkers as a novel class of biomarkers. The potential of PK markers has been overlooked. With establishment of reliable and versatile assay formats, PK markers may synergize with PD markers to deepen our understanding of treatment resistance. This figure was created with BioRender.Com.

Little information has been gathered, on the other hand, for PK of clinically approved ICBs even though the optimal function of ICBs likely requires appropriate distribution in situ. PK studies are a key component of early-phase clinical trials to ensure objective determination of a safe and effective dosing schedule. In fact, detailed PK reports are required for drug approval by the US FDA and respective authorities of other countries to ensure the safety of patients. It is important to note here that, as safety is the highest priority in clinical development, the approved dosing may not be optimally effective. Moreover, early-phase clinical trials typically involve relatively small cohorts, which are less likely to sufficiently capture PK variability across a diverse population. Recently, we and others demonstrated that patients with higher serum trough levels of ICB were more likely to achieve positive clinical responses and outcomes (Basak et al. 2019; Curti et al. 2021; Koguchi et al. 2021). These data support the usefulness of peripheral levels of mAb as a biomarker but also raise the question as to whether ICB dosing should be adjusted based on serum levels achieved in individual patients, similar to what has been attempted for the treatment of inflammatory diseases (Oude Munnink et al. 2016).

In this review, we will discuss the potential for measuring trough levels of mAbs in oncology as a clinical biomarker, mechanisms affecting mAb PK variability among patients, the potential utility of therapeutic drug monitoring (TDM) for mAbs in oncology, and the challenges and potential solutions for implementing TDM in patient care.

Peripheral therapeutic antibody levels as a novel class of on-treatment biomarker for cancer immunotherapy

Pre-treatment biomarkers may enable effective selection of treatment for patients with cancer. Pre-treatment biomarkers consist of two different categories: predictive and prognostic (Ballman 2015). For example, tissue expression of PD-L1 is a predictive biomarker for anti-PD-1/PD-L1 treatment (Topalian et al. 2012). Another example is MSI-H or dMMR status of tumors for anti-PD-1/PD-L1 treatment (Andre et al. 2020). A predictive biomarker can be used to stratify patients who are likely to respond to mAb and may limit unnecessary exposure and adverse events. Although targeting a receptor-ligand pair is straightforward way to identify a potential predictive biomarker, like in the case of PD-L1 for anti-PD-1, this tactic does not always yield a useful biomarker. For instance, the expression of CTLA-4 and its ligand CD28 are not predictive biomarkers for clinical response to ipilimumab, an anti-CTLA-4 mAb. Alternatively, prognostic biomarkers are not treatment-specific and are an indiscriminatory biomarker that tends to be associated with disease progression such as tumor burden (e.g., LDH) or physical condition of patients (e.g., ECOG status, albumin level, absolute lymphocyte count, etc.). Many biomarkers fall into this category and may not be useful clinically.

On-treatment biomarkers can provide mechanistic insights into the success and failure of mAb treatment and typically consist of PD markers. We argue that PK biomarkers including trough levels of mAb are an underappreciated class of on-treatment biomarkers, are treatment-specific, and widely applicable for many clinically approved mAbs. However, mAbs are not necessarily uniform in their function and configuration. In fact, therapeutic mAbs consist of different antibody categories from chimeric to fully human mAbs, from mAbs that inhibit receptor–ligand interactions, like ICBs, to mAbs that perform targeted depletion, and mAbs consisting of different immunoglobulin G (IgG) subtypes (e.g., IgG1, IgG2, and IgG4). Given this inherent variety, herein, we aim to systematically summarize reported peripheral levels of ICB and targeted mAbs to illustrate the potential clinical usefulness of monitoring peripheral mAb levels as on-treatment biomarkers.

In this review, we frequently use two related but distinct concepts in PK terminology: 1) Trough concentration, which is the concentration of a drug in the blood immediately before the next dose is administered (Figure 2) (Merry and Flexner 2008). When an mAb is administered in a multiple-dosing regimen, each successive dosage is administered before the preceding dose is eliminated. Accumulation of the mAb results in a higher peripheral drug concentration, which eventually plateaus where the same maximum and minimum (trough levels) concentrations are reproduced repeatedly. This state is called a steady state. In contrast, the state following administration of the first dosage is called the initial state (Figure 2); and then 2) Clearance (CL), which is an index of the ability of the body to eliminate a drug. Rather than describing the amount of drug eliminated, CL describes the volume of plasma from which a drug would be totally removed per unit time and is inversely related to half-life (Bardal et al. 2011). Therefore, low CL results in high systemic exposure (high trough levels), while high CL results in low systemic exposure (low trough levels) (Figure 3). Below, we will summarize how CL, exposure, and trough level of mAbs differ between responders and non-responders. We will also discuss the implications of varying PK using a concept of various exposure-response (E-R) relationships.

Figure 2. PK of multiple dosing of the mAb. If mAbs are administered at fixed doses and intervals, mAb accumulation occurs if intake exceeds elimination. The steady state is defined by the state where drug concentration reaches plateau so that the maximum and minimum concentrations are the same. The initial state is the state following the first administration of mAb until the second dose is administered. This figure was created with BioRender.Com.

Figure 3. Baseline-driven and response-driven E-R relationships. Baseline-driven E-R relationships are caused by baseline conditions that affect the initial CL of mAbs. Response-driven E-R relationships are caused by on-treatment conditions in response to therapeutic mAbs, which typically becomes apparent at steady state by affecting the steady state CL.

ICB: anti-CTLA-4

Ipilimumab is a fully human IgG1 mAb that targets CTLA-4 (CD152), a negative regulator of T cell activation (Wei et al. 2018). Ipilimumab blocks the interaction between CTLA-4 and CD28 to augment priming of naive T cells and was the first ICB approved by the US FDA (Lipson and Drake 2011). As a monotherapy, ipilimumab is approved for the treatment of advanced melanoma as well as in the adjuvant setting after surgical resection. In combination with nivolumab (anti-PD-1), ipilimumab is approved for advanced melanoma, renal cell carcinoma (RCC), colorectal cancer (CRC) with microsatellite instability (MSI)-high status, hepatocellular carcinoma (HCC), non-small cell lung cancer (NSCLC), mesothelioma, and esophageal cancer (EC). The population PK study of ipilimumab monotherapy (n = 498) revealed a weak but noticeable E-R relationship (Feng et al. 2013). However, most PK data are from subjects treated with 10 mg/kg, although the approved dose for advanced disease is 3 mg/kg.

In order to ask whether trough levels of ipilimumab could be used as a biomarker for clinical outcomes of patients with advanced melanoma at the approved 3 mg/kg dose, we investigated the relationship between trough levels of ipilimumab and clinical outcomes in patients with advanced melanoma (n = 37) (Koguchi et al. 2021). We found that trough levels of ipilimumab were higher in patients who developed immune-related adverse events but did not differ based on the presence or absence of disease progression. As some patients with progressive disease (PD) developed a mixed response (mixed), we included them in a separate cohort of patients without PD and then compared them with the remaining patients with PD. This analysis revealed higher trough levels of ipilimumab in the groups that achieved mixed/stable disease (SD)/partial response (PR)/complete response (CR). We also found that patients with higher trough levels of ipilimumab had better OS when grouped based on ipilimumab trough levels.

Tremelimumab is a fully human IgG2 mAb that targets CTLA-4. Although currently not approved for clinical use, tremelimumab in combination with durvalumab was granted a priority review for treating first-line HCC. Population PK analysis showed that the median OS for the 147 patients in the fast-CL group (above median CL value) was 9.6 months versus 15.8 months for the 146 patients in the slow-CL group (below median CL value) (HR = 0.54; P < 0.001) (Wang et al. 2014).

ICB: anti-PD-1

Nivolumab is a fully human IgG4 anti-PD-1 mAb that was first approved in 2014 by the US FDA and is approved for the treatment of melanoma, NSCLC, mesothelioma, RCC, Hodgkin lymphoma, head and neck squamous cell carcinoma (HNSCC), urothelial carcinoma, MSI high CRC, HCC, and EC. Investigation of E-R relationship in patients (n = 76) with NSCLC treated with second-line nivolumab therapy at 3 mg/kg Q2W demonstrated that patients with PR had higher trough levels at weeks 2, 4, and 10 than those with PD (Basak et al. 2019). Survival analysis showed that patients with higher trough levels (above median) of nivolumab achieved longer OS than those with lower trough levels. A prospective cohort study on the PK of nivolumab found that a CL-response relationship was observed in NSCLC (n = 154) and that patients with lower CL of nivolumab showed longer OS and progression-free survival (PFS) (Hurkmans et al. 2019). Results for patients with melanoma and RCC were inconclusive due to small sample size. Although not statistically significant, another study showed a trend that nivolumab-treated NSCLC patients with higher trough level at day 28 had better OS (p = 0.078, n = 68) (Bellesoeur et al. 2019).

Pembrolizumab is a humanized IgG4 anti-PD-1 mAb that was first approved in 2014. The indication of pembrolizumab includes melanoma, NSCLC, small cell lung cancer, HNSCC, Hodgkin lymphoma, primary mediastinal large B-cell lymphoma, urothelial carcinoma, MSI-high cancer and CRC, gastric cancer, esophageal cancer, cervical cancer, HCC, RCC, Merkel cell carcinoma (MCC), endometrial carcinoma, tumor mutational burden high cancer, cutaneous squamous cell carcinoma, and triple-negative breast cancer. A population PK study showed that patients with slower CL of pembrolizumab had better OS in both melanoma (n = 211) and NSCLC (n = 537) cohorts (Turner et al. 2018). Another population PK study showed that initial CL of pembrolizumab was comparable among responders and non-responders, but reduced steady-state CL as compared to initial CL was only detected in responders in both melanoma (n = 1,612) and NSCLC (1,207) (Li et al. 2017). When we analyzed trough levels of pembrolizumab in patients with melanoma and HNSCC (n = 30), patients with disease control (SD/PR/CR) at day 85 post-treatment showed higher steady state trough levels of pembrolizumab (Curti et al. 2021). We further showed that higher pembrolizumab trough levels at day 43 correlated with significantly improved PFS and OS. Similarly, patients with high pembrolizumab trough levels experienced substantially longer median OS than those with low pembrolizumab trough levels (n = 28) (Navani et al. 2021).

Anti-PD-L1

Avelumab is a fully human IgG1 anti-PD-L1 mAb that is approved for the treatment of metastatic MCC, advanced or metastatic urothelial carcinoma, and RCC. The magnitude of reduction in CL from initial to steady state was observed to be higher in responders than in non-responders in patients with MCC (Wilkins et al. 2019). Data from the US FDA Center for Drug Evaluation and Research showed high trough concentrations of avelumab were associated with prolonged OS in patients with urothelial carcinoma (Application # 761078Orig1s00).

Atezolizumab is a humanized IgG1 anti-PD-L1 mAb that was first FDA approved in 2016 and is used for the treatment of advanced or metastatic urothelial carcinoma, NSCLC, small cell lung carcinoma, HCC, and melanoma. Data from the US FDA Clinical Pharmacology Review showed that CL tends to decrease more from initial to steady state in responders, and patients with greater reduction in CL showed better OS than those with milder or no CL reduction in patients with NSCLC (Application # 761041Orig1s00). However, E-R analyses did not identify statistically significant relationships between objective response and CL in patients with urothelial carcinoma (Stroh et al. 2017).

Targeting mAbs

Rituximab is a chimeric mouse-human IgG1 that targets CD20 molecule on B cells and is approved to treat refractory B cell non-Hodgkin lymphoma, chronic lymphocytic leukemia, rheumatoid arthritis, granulomatosis with polyangiitis, microscopic polyangiitis, and pemphigus vulgaris. It depletes CD20-expressing malignant and non-malignant B cells by causing complement-dependent cytotoxicity (CDC) and antibody dependent cellular cytotoxicity (ADCC). Data from a multicenter phase III clinical trial of rituximab, which enrolled 166 patients with recurrent low-grade lymphoma, showed that the median serum level of rituximab was higher in responders than in non-responders (Berinstein et al. 1998). These differences reached statistical significance by the second infusion for pre-infusion levels and by the fourth infusion for post-infusion levels. The levels of rituximab were statistically significantly higher in the responders than in the non-responders at all three post-treatment time points. A phase II study of rituximab for relapsed patients with indolent B-cell lymphoma and mantle cell lymphoma showed significantly longer progression-free survival (PFS) in patients with higher trough levels of rituximab (n = 34) than in those with lower trough levels (n = 32) when rituximab levels were obtained before the third infusion (Igarashi et al. 2002). However, they did not observe a correlation between overall response rate (ORR) and serum rituximab levels. Another report with a smaller number of patients with aggressive B cell lymphoma showed higher trough levels of rituximab in responders than non-responders (Tobinai et al. 2004).

Alemtuzumab is a humanized IgG1 antibody that recognizes the CD52 antigen for the treatment of chronic lymphocytic leukemia. CD52, a small glycosylphosphatidylinositol-anchored glycoprotein, is highly expressed on normal T- and B-lymphocytes and on a large proportion of lymphoid cell malignancies but not on hematopoietic progenitor cells (Dumont 2002). A study with human CD52 transgenic mice showed that depletion of lymphocytes was largely mediated by neutrophils and NK cells through ADCC but not through CDC (Hu et al. 2009). Analysis from an open-label phase II clinical trial of alemtuzumab showed a significant association between high trough concentrations and improved clinical response (Hale et al. 2004). Eight responders who reached minimal residual disease (MRD) had significantly higher mean trough concentrations of alemtuzumab throughout the whole course of treatment from week 2 to week 11 than non-responders. A population PK study from four clinical trials demonstrated that the mean trough concentration of alemtuzumab was higher in responders than in non-responders (Mould et al. 2007). In the case of subcutaneous administration of alemtuzumab for patients with chronic lymphocytic leukemia, responders (n = 19) had higher median plasma concentrations than non-responders (n = 6) (Wierda et al. 2011).

Cetuximab is a recombinant mouse-human chimeric IgG1 anti-epidermal growth factor receptor (EGFR) monoclonal antibody that is indicated for KRAS wild-type, EGFR expressing CRC, BRAF V600E mutation-positive metastatic CRC, and advanced HNSCC. Cetuximab binds EGFR with higher affinity than the natural ligand EGF. Therefore, cetuximab is considered to block signals mediated by EGF-EGFR interaction and/or to induce ADCC (Garcia-Foncillas et al. 2019). In a phase I study of cetuximab in 33 patients with solid tumors, patients with clinical benefit (PR+SD) had a higher cetuximab trough concentration compared to patients with progressive disease (Fracasso et al. 2007). In 71 patients with metastatic colorectal cancer treated with cetuximab, the median PFS was significantly longer in patients with higher cetuximab trough concentrations than that of lower trough concentrations (7.8 months versus 3.3 months) (Azzopardi et al. 2011).

Although there are some minor differences between the different mAbs, taken together, these data suggest that mAb trough levels are a promising candidate biomarker for cancer immunotherapy.

Interpretation of (E-R) relationships and underlying mechanisms of varying PK

Most observed E-R relationships were not exposure-driven

Do these observed E-R relationships suggest that poor clinical outcomes are due to suboptimal exposure to mAbs? If so, can increased exposure improve the efficacy of these mAbs? Caution needs to be taken when interpreting E-R relationships as there are at least three potential scenarios for the E-R relationship, namely exposure-driven E-R, baseline-driven E-R, and response-driven E-R (Badawi et al. 2019; Dai et al. 2020). Exposure-driven E-R is the true E-R relationship where response is dictated by dose/exposure. In this case, increased dosing may improve clinical outcome. Baseline-driven E-R considers situations where confounding baseline factors influence both exposure and response. In other words, baseline factors dictate both CL of mAbs and clinical outcome, but independently. For example, these baseline factors could include tumor burden, poor ECOG performance status, and/or catabolic state. Finally, Response-driven E-R relationships refer to a response to the treatment that alters CL and exposure over time. This typically happens at a steady state level when treatment-mediated reduction in tumor burden can decrease target-mediated drug disposition (TMDD) in responders but not in non-responders (Figure 3).

Analysis of clinical trials with different dosing levels can clarify whether the observed E-R relationship is truly exposure-driven E-R. The population PK study of pembrolizumab analyzed OS from the 2 mg/kg and 10 mg/kg arms (Turner et al. 2018). When they compared OS among patients in the 1^st (slow CL) versus 4^th quartile (rapid CL) based on CL of pembrolizumab, they discovered that subjects in the 1^st quartile from the 2 mg/kg cohort demonstrated substantially better OS than those from the 4^th quartile in the 10 mg/kg cohort. The authors also discovered that subjects in the 4^th quartile for CL from the 2 mg/kg cohort exhibited very similar OS as the 4^th quartile for CL in the 10 mg/kg cohort. In other words, it was CL, but not exposure (or trough levels), that correlated with clinical outcome.

Another study evaluated E-R relationships across three different doses of nivolumab for patients with advanced melanoma (Agrawal et al. 2016). This study demonstrated that higher trough levels correlated with increased clinical response to nivolumab in each of the 1, 3, and 10 mg/kg dosing groups. However, even though subjects with lower trough levels in the 10 mg/kg group achieved much higher trough nivolumab levels compared to the 1 and 3 mg/kg groups, those patients did not respond to nivolumab. Importantly, this study revealed almost overlapping Hill function sigmoid curves between the probability of response and CL of nivolumab among three different doses (slow CL, better response).

On the other hand, a population PK study of ipilimumab showed weak but potential exposure-driven E-R (Feng et al. 2013). Patients who received 10 mg/kg of ipilimumab showed slightly better clinical outcomes than those with 0.3 mg/kg or 3 mg/kg dosing. Unfortunately, there were significantly higher incidences of adverse events with 10 mg/kg dosing, which precludes the clinical utility of this dose.

Together, population PK studies comparing multiple dosing levels are indispensable for clarifying the presence of true exposure-driven E-R and for adjusting dosing if necessary.

Baseline-driven E-R relationships

In our previous work, we identified baseline, but not on-treatment, CXCL11 as a predictive biomarker that was negatively associated with clinical outcome in patients with advanced melanoma treated with ipilimumab but not in patients treated with a gp100 vaccine (Koguchi et al. 2015). We hypothesized that baseline CXCL11 reflects pre-treatment conditions affecting clinical outcome and trough levels of ipilimumab, but independently. To address this hypothesis, we conducted a multivariate analysis including trough levels of ipilimumab and CXCL11 along with other demographic information as confounding factors to assess the association of both factors with OS. When we used trough levels of ipilimumab as a categorical valuable (below or above median), CXCL11 lost its association with clinical outcome. On the other hand, when we used trough levels of ipilimumab as a continuous valuable, then trough levels of ipilimumab, but not CXCL11, lost association with clinical outcome (Koguchi et al. 2021). These results suggested that baseline conditions are likely responsible for elevated CXCL11 levels and lower trough levels of ipilimumab.

In contrast, expression of the predictive biomarker PD-L1 did not have any effect on initial CL of nivolumab (Bajaj et al. 2017) and pembrolizumab (Turner et al. 2018). This is surprising, but PD-L1 expression may have a larger effect on steady state CL than initial CL of ICB (Wilkins et al. 2019). Instead, Bajaj et al. found that baseline albumin and LDH had a significant effect on CL (>20% and within 20%, respectively) and lower albumin and higher LHD levels had higher CL (Bajaj et al. 2017). Thus, for PD-1/PD-L1 baseline conditions represented by prognostic biomarkers such as decreased albumin and elevated LDH have a significant effect on CL, which suggests involvement of cachexic conditions and higher tumor burden, respectively.

Response-driven E-R relationships and its proposed underlying mechanisms

It is conceivable that some cancer-specific common conditions affect PK and subsequently manifest as response-driven E-R since E-R relationships typically become evident at steady state when the patient’s response in terms of changes in tumor burden can be clinically evaluated. In other words, responders are liberated from rapid CL caused by tumor-associated factors, while non-responders continue to experience rapid CL of mAbs. Inflammation is a common feature associated with tumorigenesis and cancer progression (Greten and Grivennikov 2019) that may also affect PK in cancer patients. Examples of inflammation affecting antibody levels exist in patients with chronic inflammatory conditions, such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA), where IgG, but not IgM, catabolism is increased (Levy et al. 1970; Wochner 1970). Notably, this increase in IgG catabolism cannot be explained by increased loss of IgG in the kidney or intestines or by elevated circulating IgG levels. In a preclinical model, we showed that acute and systemic inflammation led to accelerated CL of mAbs (Laws et al. 2016), suggesting that inflammation may influence mAb trough levels in humans. Therefore, we examined the relationship between on-treatment levels of inflammation (CRP and IL-6) and trough levels of ipilimumab, and found that they were inversely associated (Koguchi et al. 2021). Another condition common in advanced cancer is a hypercatabolic state called cachexia (Baracos et al. 2018). Cachexia manifests with specific loss of skeletal muscle and adipose tissue through a combination of reduced food intake and metabolic changes, including elevated energy expenditure, excess catabolism, and inflammation. The population PK study of pembrolizumab found an association between heightened CL of pembrolizumab and the presence of cachexia (judged by body weight and albumin levels) (Turner et al. 2018). Although this was shown in relation to initial CL, we envision this is relevant to response-driven E-R as non-responders would continue to experience the negative effects of cachexia. Preclinical models showed that tumor-derived IL-6 causes metabolic changes that result in cachexia. Although we did not observe changes in body weight and albumin levels in our cohort of patients with melanoma treated by ipilimumab, subjects with elevated IL-6 levels may have been in the pre-cachexia condition (Flint et al. 2016).

The level of mAbs is tightly regulated by the balance between a) recycling of mAbs by neonatal Fc receptors (FcRn) that protect mAbs from lysosome-mediated antibody disposition and b) disposition by proteolysis in lysosomes of endothelial and monocytic cells (Liu 2018). Due to this relationship, downregulation of FcRn, which results in reduced mAb recycling, may be responsible for increased CL of mAbs in patients with high tumor burden. This would be akin to the relationship between FcRn and albumin, as FcRn also recycles albumin. Furthermore, patients with FcRn-low NSCLC were associated with a poor prognosis, probably due to a metabolic advantage through high consumption of albumin (Dalloneau et al. 2016). IFN-γ is known to downregulate FcRn expression as a tumor extrinsic factor (Liu et al. 2008) and therefore may contribute to decreased mAb trough levels. How other pro-inflammatory markers, such as IL-6 and/or CRP, modulate FcRn expression is still largely unknown. Taken together, we envision that tumor intrinsic and extrinsic factors could downregulate FcRn and increase disposition of mAbs, both of which could result in increased CL. On the other hand, elevated IL-6 and CRP may simply correlate with increased systemic inflammation leading to increased CL and decreased trough levels of mAbs. Further mechanistic studies are needed to help clarify what is causal and what are downstream events in this context.

Response-driven E-R is typically observed when mAbs achieve a steady state drug level. This typically manifests as time-varying PK where mAb CL drops significantly when the mAb concentration reaches steady state. It is believed that treatment response changes CL through reducing the catabolism of mAb by resolving inflammation status in responders (Dai et al. 2020). Another factor that can contribute to response-driven E-R is target-mediated drug disposition (TMDD). In this case, mAb CL is believed to occur in saturable target-mediated mechanisms such as receptor-mediated endocytosis and subsequent lysosomal degradation of the mAb-receptor complex. At low mAb concentrations, total CL is relatively unaffected, and target-mediated elimination represents a major elimination pathway. With increased mAb concentrations, the target-mediated elimination pathway starts to become saturated. As a result, total CL decreases substantially (An 2020).

Dynamic and circular interactions between drug exposure and clinical response

Although ipilimumab PK was originally reported to be time-invariant (Feng et al. 2014), recent work showed that ipilimumab CL decreased over time and the change in CL was greater in patients treated with nivolumab combination than ipilimumab alone and in responders than non-responders (Sanghavi et al. 2020). Importantly, interactions between exposure and response are circular. While an adequate mAb concentration results in disease control or remission, a subtherapeutic mAb concentration could result in disease progression. Disease activity then directly affects mAb target expression. Target-mediated drug disposition results in low mAb concentrations in non-responders, while adequate concentrations can be maintained in responders (Dai et al. 2020; Oude Munnink et al. 2016).

Therapeutic drug monitoring for ICB

The presence of an E-R relationship, especially exposure-driven E-R, suggests a potential opportunity to adjust dosing to optimize mAb therapy. This has been most explored in the field of anti-TNF-α therapy, especially infliximab, a chimeric antibody approved by the US FDA in 1998 (Papamichael et al. 2019). The use of infliximab significantly improves symptoms of inflammatory bowel diseases (IBDs). However, a significant proportion of patients experience either a primary failure or a secondary loss of response (LOR) during treatment. Based on this notion, therapeutic drug monitoring (TDM) was introduced to reduce the incidence of LOR by maintaining effective mAb exposure. TDM can be driven by clinical episode (reactive TDM) or scheduled to detect the earliest sign of LOR (proactive TDM) (Shmais et al. 2021). Although dose adjustment based on TDM is still controversial and challenging to implement, improved TDM (e.g., point of care TDM) may facilitate wider implementation and potential benefit particularly for patients with IBDs.

It remains unclear whether implementing TDM will improve the efficacy of oncology-specific mAbs; however, TDM may provide advantageous information if it meets certain criteria (Oude Munnink et al. 2016). First, the exposure-driven E-R relationship and target range of trough concentrations need to be firmly established for each specific mAb and indication. Second, the optimal timing and scope of mAb-TDM should be determined (e.g., at treatment induction, intermittent during treatment maintenance, and/or at loss of response). Third, prospective randomized trials are needed to show mAb-TDM-guided clinical decision-making superiority above standard of care therapy. Fourth, treatment algorithms for TDM-guided clinical decision-making should be developed. As discussed above, most of the observed E-R relationships are baseline or response-driven. Thus, TDM may not be an appropriate avenue for improving clinical outcome. On the other hand, TDM might facilitate cost-effective use of mAbs by allowing an option of increasing the interval of mAb dosing (Chatelut et al. 2021).

Assay considerations

Finally, there is a critical need for a reliable analytical technique applicable to assay the ever-increasing number of therapeutic mAbs. The enzyme-linked immunosorbent assay (ELISA) is typically used for measuring mAbs. However, ELISAs tend to suffer from poor sensitivity/selectivity at lower concentrations, non-specific binding as it is an indirect technique, interference from anti-drug Abs, and lack of multiplexing capability. Even though planar and solution-based multiplexing applications are available, they suffer from lack of adequate standardization and quality control (Tighe et al. 2015). In contrast, liquid chromatography-mass spectrometry (LC-MS)-based mAb assays directly detect a structure unique to each mAb and therefore are compatible with multiplex quantitation (Iwamoto and Shimada 2018; Todoroki et al. 2020). However, typical LC-MS-based mAb assays have been challenged with analytical instability due to excess tryptic peptides and autolysis of trypsin enzyme during whole analyte digestion.

Iwamoto et al. developed nano-surface and molecular-orientation limited (nSMOL) proteolysis to overcome such challenges (Iwamoto et al. 2014). The nSMOL assay captures IgGs using a Protein A resin with a 100 nm pore to orient their Fab to the reaction solution. IgGs are then proteolyzed by trypsin immobilized on the surface of nanoparticles with a 200 nm diameter. Immobilized trypsin has physicochemically limited access to the Fab of IgGs because of the difference of the two resins, therefore decreasing the peptide number while maintaining the structural specificity of complementarity-determining region peptides for downstream multiple reaction monitoring with triple-quadrupole LC-MS. Currently, nSMOL can be used to identify signature peptides of more than 30 unique mAbs (Iwamoto and Shimada 2022). We recently developed an innovative assay based on nSMOL technology that could accurately measure concentrations of different mAbs using one universal reference antibody, named refmAb-Q (Iwamoto et al. 2022). This may greatly facilitate the implementation of the assay at a central laboratory.

Concluding remarks

Currently, 111 mAbs have been approved by the US FDA, 2 mAbs are approved under emergency use only (anti-SARS-Cov2 mAbs), and 22 mAbs are currently in review as of mid-August 2022 (https://www.antibodysociety.org/resources/approved-antibodies/). This review highlights the fact that mAbs can be viewed as a class of therapeutics and can also be used as a biomarker for clinical outcome. Levels of mAbs may also provide insight into the inflammatory status of the TME if we can further understand the link between TME-specific inflammation and mAb levels. Drug monitoring during clinical trials may help address this issue, followed by a prospective study to validate these findings including the appropriateness of a cutoff value. PK information collected during the primary mAb treatment may inform the strategy for secondary mAb treatment if the patient does not respond to the primary mAb. With appropriate standardization and validation of the detection assay, it might be possible to incorporate this approach into the clinical workflow to support rational mAb treatment. This has the potential to improve patient outcomes by offering more personalized treatment while reducing costs and potential adverse events associated with ineffective therapies.

Disclosure statement

YK: research support from Bristol Myers Squibb (BMS), GlaxoSmithKline, and Shimadzu. WLR: research support from BMS, GlaxoSmithKline, MiNA Therapeutics, Inhibrx, Veana Therapeutics, Shimadzu, OncoSec, Galecto, Canwell Pharma, Turn Bio, and Calibr; patents/licensing fees: Galectin Therapeutics; advisory boards: Medicenna, Vesselon, and Veana Therapeutics.

Additional information

Funding

The work was supported by the Shimadzu Providence Portland Medical Foundation .