A new frontier in haematology – combining pharmacokinetic with pharmacodynamic factors to improve choice and dose of drug

Abstract

The issue of tailored dosing adjusted according to a range of patient-specific factors other than bodyweight or body surface area is of large and increasing clinical and financial concern. Even if it is known that dosing alterations are likely to be required for parameters such as body composition, gender and pharmacogenetics, the amount of dosing change is unknown. Thus, pharmacokinetically guided dosing is making a resurgence, particularly in areas of medicine where there are cost constraints or safety issues, such as in haematology medications. However, the evidence to support the behaviour is minimal, particularly when long-term outcomes are considered. In haematology, there are particular issues around efficacy, toxicity and overall cost. Newer targeted agents, such as the monoclonal antibody rituximab and the tyrosine kinase inhibitor imatinib, whilst clearly being highly effective, are dosed on a milligram per square metre (rituximab) or fixed dose basis (imatinib), regardless of body composition, tumour aspects or comorbidity. This review questions this practice and raises important clinical issues; specifically, the clinical potential for combined pharmacokinetically and pharmacodynamically guided dosing of new targeted agents in haematological malignancies. This pharmacokinetically and pharmacodynamically guided dosing is an emerging area of clinical pharmacology, driven predominantly by toxicity, efficacy and cost issues, but also because reasonable outcomes are being noted with more appropriately dosed older medications adjusted for patient-specific factors. Clinical trials to investigate the optimization of rituximab dose scheduling are required.

Introduction

The choice and use of therapeutics in haematological malignancies is characterized by knowledge from two different eras. The serendipitous discovery of the effects of mustard gas on the bone marrow and lymph node during World War II led to the introduction of nitrogen mustard as a treatment for a patient with non-Hodgkin's lymphoma (NHL) [1]. This enabled numerous other programmes devoted to the growth of other drugs with similar effects to be developed. Subsequently, in an attempt to improve efficacy and reduce toxicity, pharmacokinetic studies were undertaken. Specifically, concentration-dependent survival and toxicity were demonstrated with several drugs, including methotrexate and busulfan. Evans et al. showed that a concentration–effect relationship exists for high-dose methotrexate when administered in patients with acute lymphocytic leukaemia [2]. This led to more patient-specific dosing regimens in haematology.

More recently, highly targeted agents, such as the monoclonal antibody (mAb) rituximab, used in B-cell NHL, and the small molecular inhibitor imatinib, used in chronic myeloid leukaemia (CML), have been developed. The use of such highly efficacious agents that exhibit relatively modest toxicity has radically redefined clinical decision making in haematological malignancies. Current research is suggesting that better patient outcomes may be achieved if such sophisticated agents are delivered at specific doses to achieve certain concentrations, after taking into account both patient- (i.e. body size and composition) and tumour-specific factors. This review provides an overview of this move towards individualization of choice and dose of haematology drugs. Whilst the review focuses on the anti-CD20 mAb rituximab (using imatinib for comparison), the principles are likely to be broadly applicable to other haematological medications.

Rituximab

Rituximab (MabThera®, Rituxan®) is a chimeric monoclonal antibody genetically engineered to fuse the light- and heavy-chain variable regions of 2B8, a murine monoclonal antibody, with the human κ light-chain and IgG1 heavy-chain constant regions [3,4]. Rituximab does not induce human antimouse antibody responses, and human antichimeric responses occur in only 1% of patients and are transient and not clinically significant. It is targeted against CD20, a transmembrane antigen expressed on both pre-B and mature B lymphocytes, which has been shown to be involved in B-cell activation, regulation of B-cell growth and the control of calcium handling [5,6]. Although in vitro studies have demonstrated rituximab-induced apoptosis, complement-mediated cytotoxicity, antibody-dependent cellular cytotoxicity and Fcγ receptor 2/CD32-dependent phagocytosis, the relevance and contribution of each mechanism of action to the clinical response of patients is yet to be elucidated [3,7,8].

CD20 is homogeneously expressed in over 90% of B-cell lymphomas and chronic lymphocytic leukaemia (CLL), and over a relatively brief period rituximab has become part of the standard of care for CD20⁺ lymphoproliferative disorders; these include follicular lymphoma (FL; the most frequent indolent B-cell NHL), diffuse large B-cell lymphoma (DLBCL; the commonest aggressive B-cell NHL) and CLL (the highest-incidence adult leukaemia). In a seminal early trial involving elderly patients diagnosed with DLBCL, the use of rituximab in conjunction with CHOP chemotherapy (cyclophosphamide, doxorubicin, vincristine and prednisolone) resulted in higher response rates and improved event-free survival and overall survival than with CHOP alone [9]. Similar clinical benefits with the addition of rituximab to CHOP combination chemotherapy in younger patients with DLBCL have also been observed [10]. A number of phase III studies have also demonstrated the superiority of rituximab in combination with chemotherapy for the treatment of symptomatic stage III and IV FL in the front-line and relapsed/refractory settings [11]. The role of rituximab monotherapy in asymptomatic FL is currently under evaluation, whereas maintenance single-agent rituximab is well established as being beneficial in terms of overall survival for FL [12]. For CLL, two recent randomized clinical trials indicate that the addition of rituximab to FC chemotherapy (fludarabine and cyclosphosphamide) enhances response rates and prolongs progression-free survival (PFS) compared with FC alone in previously untreated and refractory/relapsed patients [13,14]. Non-neoplastic haematological diseases in which rituximab has shown activity include autoimmune disorders such as immune thrombocytopenia, as well as nonhaematological diseases such as rheumatoid arthritis [15,16].

The recommended dose of rituximab administered as a single agent in patients with indolent B-cell non-Hodgkin's lymphomas (B-NHLs) is 375 mg m⁻² given weekly for 4 weeks. Likewise, when co-administered with CHOP chemotherapy, the dose is 375 mg m⁻² with each cycle. However, it must be emphasized that these decisions are based on empirical considerations [3]. Notably, several studies indicate wide interindividual variation in rituximab serum concentrations, which is important because a relationship exists between response and mAb levels [17–19]. Furthermore, although rituximab is generally well tolerated, a dose-escalation study in CLL (at doses ranging from 500 to 2250 mg m⁻²) reported a significant increase in infusion-related toxicities [20]. Based on these observations, it has been proposed that the ‘375’ dose regimen could be optimized by adjustment for patient-specific factors [3]. However, the amount of dose change in particular cases is not known.

Factors affecting rituximab exposure

The pharmacokinetic profile of rituximab follows that of the two-compartmental model, with the mean distribution and elimination half-lives being approximately 1.3 and 19 days, respectively [3]. However, there is large interindividual variability in these parameters of rituximab [21]. It is likely that variability relates to both tumour-related factors (antigen density on the malignant B cell and extent of tumour burden) but also host genetics, gender, bodyweight and dosing frequency (Table 1 ).

Table 1.

Summary of studies of host factors affecting rituximab exposure

Factor	Study references	Key findings
FCGR3A-158V/F gene polymorphism	Dall'Ozzo et al. (2004) [7]	FCGR3A-158V polymorphism has higher affinity for rituximab compared with the FCGR3A-158F polymorphism.
		Lower EC50 with natural killer cell-mediated antibody-dependent cellular cytotoxicity with the homozygous V genotype than with the homozygous F genotype
	Ternant et al. (2012) [28]	In silico model predicted that 1500 mg m−2 maintenance doses of rituximab confer benefit in patients with FL, and that FCGR3A-158F carriers will have lower PFS rates compared with homozygous V genotype patients, even with higher rituximab doses
Gender	Müller et al. (2012) [32]	Rituximab clearance in elderly male DLBCL patients was significantly faster than in elderly female patients
	Ng et al. (2005) [16]	Gender was found to be a greater factor for interindividual variability than body surface area
Weight	Müller et al. (2012) [32]	Faster rituximab clearance is associated with higher bodyweight in DLBCL patients.
		Significant improvement in PFS was identified in patients with bodyweight in the lower quartile for their gender
Dosing frequency	Pfreundschuh et al. (2011) [41]	SMARTE-R-CHOP-14 was designed to give higher initial doses and prolong duration of rituximab exposure with the final doses. This resulted in early achievement of maximal trough levels for rituximab and prolonged duration of detectable serum rituximab levels
	Delarue et al. (2009) [38]; Cunningham et al. (2011) [39]	Equivalent efficacy in patients with DLBCL treated with R-CHOP-21 × 8 compared with R-CHOP-14 × 6 and ×2R

Abbreviations are as follows: CHOP, cyclophosphamide, doxorubicin, vincristine and prednisolone; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; PFS, progression-free survival; R, rituximab.

B-Cell antigenic density

In addition to antibody specificity, B-cell antigenic density has long been noted to be associated with the response to rituximab [22]. More recently, antigenic density and loss of CD20 expression have been studied to improve treatments for resistance to rituximab [23]. An additional research line has focused on identifying a CD20 antigen surface threshold level required for effective rituximab-associated complement-mediated cytotoxicity. Although a direct correlation between CD20 surface expression and rituximab-associated complement-mediated cytotoxicity has been observed only in rituximab-sensitive cell lines to date [24], there is ongoing work in this area. Identification of a threshold is ideal data for building a therapeutic dosing model for rituximab in lymphoma.

Tumour density and maximal tumour diameter

Studies have shown that tumour burden per se is also related to survival. In order to investigate this, an animal study has recently evaluated the influence of B-cell density on dose–concentration–response relationships of rituximab. In this study, murine lymphoma cells, transduced with human CD20 cDNA and transfected with luciferase plasmid, were intravenously injected into C57BL/6J mice. Using a 20 mg kg⁻¹ dose, rituximab concentrations were inversely correlated with tumour burden. Furthermore, rituximab exposure influenced response and survival [25]. In a related manner, Pfreundschuh et al. have shown that rituximab improves the adverse prognostic effect of maximal tumour diameter, which has a linear prognostic effect on outcome in DLBCL. This study was used to set a cut-off point of 10·0 cm to delineate those patients with bulky disease proposed for rituximab therapy [26].

FCGR3A gene polymorphism

The FCGR3A gene encodes for the low-affinity receptor FCGR3A/CD16 found on natural killer (NK) cells and monocytes, which binds to the IgG Fc portion. This results in the activation of antibody-dependent cell-mediated cytotoxicity of these cells [7]. It has been demonstrated that this gene has a bi-allelic polymorphism resulting in either a phenylalanine (FCGR3A-158F) or valine (FCGR3A-158V) at the amino acid position 158 [7]. Furthermore, the two receptor allotypes have been shown to have different degrees of affinity for rituximab [27]. In a study conducted by Dall'Ozzo et al., a comparison of the rituximab concentrations required to inhibit the binding of anti-CD16 monoclonal antibody showed that FCGR3A-158V has a higher affinity for a rituximab than FCGR3A-158F. Furthermore, rituximab concentrations that were required to obtain 50% lysis (EC₅₀) with NK cells homozygous for the valine allotype (V/V donors) was significantly lower than that required for F/F donors [7]. This shows that the patient's genotype for the FCGR3A gene may have an influence on the rituximab concentration–effect relationship. Increasing concentrations in F carrier patients to obtain the same response seen in patients of the FCGR3A-158V genotype may then confer some benefits of optimizing the use of rituximab in clinical practice [28].

The same Fc receptor polymorphism has been associated with the development of late-onset neutropenia (LON), a recognized late complication induced by rituximab-containing therapy [29,30]. Late-onset neutropenia is defined as an unexplained grade III–IV neutropenia that develops at least 4 weeks after the last rituximab dose. In a study on patients with newly diagnosed B-cell lymphoma, Li et al. showed an association between rituximab concentrations and immunoglobulin Fc receptor polymorphisms [29]. In a separate study involving DLBCL patients treated with the CHOP-R regimen, Keane et al. demonstrated a positive influence of FCGR3A polymorphisms on the risk of developing LON, with 50% of patients homozygous for the FCGR3A-158V genotype developing LON compared with 7% V/F patients and 2% of patients homozygous for FCGR3A-158F [31]. It is suggested that the occurrence of LON may lead to good prognostic outcomes in B-cell lymphoma patients, although lead-time bias may provide an alternative explanation [30,31].

Gender

Gender differences of B-NHL patients have also been shown to affect the pharmacokinetics and, therefore, the treatment outcomes of rituximab-containing therapies [32]. This gender effect on rituximab has previously been shown in rheumatoid arthritis patients treated with rituximab [16]. In a study conducted by Müller et al., it was shown that elderly male patients with DLBCL have significantly faster rituximab clearance than female patients, which was associated with smaller benefit with the addition of rituximab to chemotherapy regimens as analysed in the RICOVER-60 trial [32]. These findings were independent of bodyweight, and the difference in clearance between male and female patients has been mainly attributed to gender-dependent differences in the activity of hepatic enzymes, the lower cardiac output and hepatic blood flow in females compared with males [33], as well as differential distribution of the drugs into fat [34]. This might explain why analyses of newly diagnosed patients with DLBCL showed better PFS and overall survival for female than male patients in rituximab-containing regimens [35]. Increasing the rituximab dosage in male patients may therefore have better results [32,35]. Likewise, in patients with previously untreated FL receiving induction chemo-immunotherapy followed by rituximab maintenance therapy, higher trough rituximab serum levels were associated with female sex as well as with absence of initial bone marrow infiltration. Notably, trough levels correlated with response and PFS [36].

Bodyweight

Another factor that has also been shown to have an effect on the pharmacokinetics of rituximab is bodyweight [32]. The improvement in the PFS of patients with DLBCL due to the addition of rituximab was shown to be significantly greater in patients with bodyweights in the lower quartile than those in the upper quartile. As both obese and undernourished subjects are often excluded from clinical trials, data on the relevant body surface area or other measurements of body size and composition obtained from clinical data are needed to make definitive dosing statements on these groups of patients.

Dosing frequency

The combination of rituximab with CHOP chemotherapy is the gold standard of care for patients with aggressive NHL, but the optimal dose and number of cycles of rituximab given is yet to be defined [3]. Administration of combination chemotherapy every 2 weeks (‘CHOP-14’) leads to an improved outcome for elderly patients with DLBCL compared with chemotherapy given 3 weekly (‘CHOP-21’) [37]. However, with the addition of rituximab to combination chemotherapy (‘R-CHOP’), there was no superiority of R-CHOP-14 compared with R-CHOP-21 seen in two randomized trials [38,39]. The observation that with R-CHOP-14 the trough rituximab concentrations did not reach plateau until cycle four or five [40] led to concerns that a potential advantage of CHOP-14 over CHOP-21 chemotherapy might be compromised by the shorter exposure to rituximab when eight applications are given every 2 weeks (R-CHOP-14; last application day 99) compared with 3 weeks (R-CHOP-21; last application day 148). This led to a recently developed dosing strategy, called SMARTE-R-CHOP-14, which prolongs the administration of rituximab to poor-prognosis DLBCL patients, with the last few doses given on completion of chemotherapy. Higher initial doses are also given, with the aim of achieving early high serum rituximab concentrations and maintaining the concentrations longer [41], thus providing a higher overall exposure. In a phase II study, higher PFS and overall survival were shown in study participants with poorer prognosis under the SMARTE-R-CHOP-14 in comparison with the schedule used in the RICOVER-60 trial [42]. Furthermore, pharmacokinetic analysis showed that maximal rituximab trough concentrations were indeed achieved earlier and remained in the serum for a longer time. A preliminary analysis indicates that the SMARTE-R schedule may be superior to 2 weekly dosing of rituximab, in particular for patients with high tumour burden.

Rituximab and pharmacokinetic–pharmacodynamic modelling

The variability of clinical outcomes seen in B-NHL patients and their association with factors affecting rituximab exposure indicate that a relationship might exist between the concentration of and/or patient exposure to rituximab and its therapeutic effects. A pharmacokinetic–pharmacodynamic (PK–PD) in silico model has been developed to predict PFS in patients with FL receiving rituximab maintenance therapy [28]. This model accounted for the influence of FCGR3A genotypes and the interindividual variability of serum levels and was applicable to patients who received either rituximab alone as induction therapy or those who received R-CHOP. It predicts that higher maintenance doses (1500 mg m⁻²) are required, but that despite this the PFS would be lower in patients with the FCGR3A-F158 allele. To date, the model remains to be validated in prospective clinical trials.

The existence of such a concentration–effect relationship might potentially serve as the basis for therapeutic drug monitoring and target concentration intervention, i.e. modifying dose according to concentrations and response, for optimizing the use of rituximab in patients with CD20⁺ lymphoproliferative disorders [42]. Although the literature for rituximab is sparse, there have been studies demonstrating relationships between PK/PD factors of other chemotherapeutic drugs for haematological malignancies with clinical outcomes. For example, in a seminal early study of therapeutic drug monitoring in haematological malignancy, children with acute lymphoblastic leukaemia were randomized to conventional (body surface area) vs. individualized therapy (based on rates of drug clearance) for methotrexate, tenoposide and cytarabine. Adjustment of methotrexate (but not tenoposide or cytarabine) for the patient's ability to clear drug resulted in reduced risk of early relapse [43].

Comparison with imatinib

As opposed to NHL, chronic myeloid leukaemia is a relatively rare haematological malignancy. However, in-depth understanding of the pathogenesis of CML has evolved at an unprecedented rate, leading to development of the small-molecule tyrosine kinase inhibitor, imatinib mesylate (Gleevec®, Glivec®). Emerging knowledge of imatinib for the treatment of CML provides a pertinent illustration of the impact of pharmacodynamics and pharmacokinetics on newer targeted agents for haematological malignancies.

Assessment of therapeutic response with imatinib is based upon meeting defined haematological, cytogenetic and molecular chronological milestones [44]. Patients failing to achieve these are described as primarily resistant to therapy, whilst those who attain the milestones but subsequently lose response are termed secondarily resistant. Current consensus is that patients failing to meet chronological milestones to imatinib should be screened for PD- and PK-associated mechanisms of resistance and then considered for imatinib dose escalation, a second-generation tyrosine kinase inhibitor, or allogeneic haematopoietic stem cell transplantation [45].

Pharmacodynamics

The mechanism of resistance to imatinib appears to be due a variety of factors, including breakpoint cluster region-Abelson (BCR-ABL) tyrosine kinase gene amplification [46]. In a proportion of patients with inherent or acquired resistance to imatinib, there are point mutations in the Abelson tyrosine kinase domain that lead to specific single amino acid substitutions that interfere with binding of the drug to the kinase, leading to an increased IC₅₀ [47].

Pharmacokinetics

Data suggest that individual pharmacokinetic parameters are important in clinical outcomes with imatinib. Specifically, plasma trough levels >1000 ng ml⁻¹ are correlated with a better clinical response in CML patients treated with imatinib [48,49]. Genetic polymorphisms have also been shown to influence the phenotype of drug transporters and, therefore, drug concentrations in the blood and tumour cell [50]. In CML, functional studies have shown that the differential activity of influx and efflux pumps affects the uptake of imatinib into the leukaemic cells and, therefore, the response to therapy [51]. The high activity of OCT-1, an influx pump for imatinib, has been shown to result in higher response rates in CML patients compared with those patients with low activity [52]. Although the same relationship has also been shown between molecular response and mRNA levels of OCT-1, there are also studies querying the clinical relevance of the genetics and mRNA expression of OCT-1 transporters with imatinib use [53] and, as such, no firm recommendations as to whether these data add any prediction for clinical outcome in addition to BCR-ABL1 mRNA levels have been made.

Multidrug resistance gene (MDR1) polymorphisms, which encode for efflux pumps such as ABCB1, also influence patient response to therapy. Ni et al. showed that patients homozygous for the T allele at loci 1236 and 3435 had poorer response compared with those having fewer copies of the allele [54]. This finding is consistent with an earlier study demonstrating that homozygosity for the T allele at locus 1236 correlated with higher plasma levels and more frequent molecular response [55].

Future directions

Rituximab has revolutionized the treatment of B-NHL through the specific targeting of the CD20 antigen. However, as with imatinib, both patient- and tumour-related factors affect the clinical response to this targeted agent. Although simulation work using target area-under-the-curve measurements is possible, clinical research collecting information on concentrations, accurate assessments of tumour burden, body composition, FCGR3A genotypes and outcomes is necessary to validate and refine development of a dosing algorithm. This work should be undertaken within the context of well-defined prospective cohorts; however, inclusion of ‘special groups’, such as patients over 60 years old and obese, is required to harness the potential benefits of individualized dosing.

Competing Interests

DRA affirms that there are no competing interests within the submitted manuscript. MKG has received travel assistance from Roche Australia to attend an educational meeting. JHM has received consulting funds from Roche to provide generic pharmacology advice.